Projects

Project #1 – Development of Optical Imaging Technology 

Description:

Optical coherence tomography (OCT) is one of the fastest growing areas of optical imaging with many potential clinical applications. Research areas that are of particular interests include the development of optical coherence tomography (OCT), Doppler optical coherence tomography, optical coherence elastography, nonlinear optical microendoscopy, and photoacoustic tomography for imaging tissue structure and physiology. The technology will be translated to clinical applications for the diagnosis and management of cardiovascular diseases and cancers.

Students’ Involvement and Expected Outcomes:

Students selected for this research project will study essential optical imaging techniques, participate in the construction of optical imaging systems and miniature endoscopic probes, conduct experiments to collect imaging data, and develop software programs for image processing.

Prerequisites:

Students with a physics or engineering background is preferred. Experience in programming and optics is a plus.

Project #1 – Quantitative Evaluation of Bone

Description:

A number of diseases can affect bone strength and development, affecting poor bone mineralization, altered morphology, and increased risk of bone fracture at any age. Early detection and treatment of these conditions can help prevent fractures. However, current technology to evaluate bone has limited sensitivity. The goal of this research is to quantify bone mineralization and strength through analysis of quantitative computed tomography scans. These analyses may be of pediatric subjects or adults, depending on the ongoing work in the lab.

Students’ Involvement and Expected Outcomes:

Students involved in this project will learn about anatomy and biology of bone, x-ray imaging, x-ray computed tomography scans (CT or CAT scans), and quantitative computed tomography of bone. Students will perform quantitative analyses of bone in patients, will formulate and test hypotheses, and will interpret the results.

Prerequisites:

Engineering, math or physics major required. Knowledge of unix and/or computer programming is highly preferred, and would greatly enhance the experience, but is not required.

Project #2 – Spine-Rad Brachytherapy Bone Cement

Description:

Spinal metastases are a common manifestation of many cancers such as those originating in the prostate, breast, lung, kidney, and thyroid. Approximately 200,000 people with spinal metastases die each year in the United States. These metastases are painful, reduce bone strength, and can lead to vertebral collapse and serious neurological complications. Conventional treatment of vertebral metastases involves external beam radiation therapy (EBRT) to slow tumor progression and potentially alleviate pain, although EBRT can further weaken bone and lead to vertebral fracture. Vertebroplasty or kyphoplasty (percutaneous injection of bone cement into the vertebral body) can be performed prior to EBRT to restore bone strength and provide immediate marked or complete pain relief in 50% to 85% of cases. Although this approach is an improvement over EBRT alone, a major shortcoming of EBRT remains: EBRT irradiates the spinal cord, limiting the dose that can be delivered to tumors

To address the limitations of conventional treatments for vertebral body metastases, UCI researchers have developed Spine-RadTM Brachytherapy Bone Cement which consists of an FDA-approved bone cement that has been mixed with an insoluble radioactive powder, P-32-hydroxyapatite (P-32-HA). Spine-Rad Cement would be delivered percutaneously (through a needle passing through the skin) to the vertebral body to restore bone strength and provide pain relief while simultaneously providing local radiation (brachytherapy) to the tumors. Dosimetry studies have shown that, because P-32 is a beta emitter, a high dose can be delivered to tumors nearby while virtually eliminating radiation to the spinal cord. This single procedure would also be more convenient for patients than multiple EBRT treatment sessions and would cause fewer side effects because radiation would travel only a short distance in the body.

Students’ Involvement and Expected Outcomes:

A student working on this project would have the opportunity to be involved in a range of activities related to radiation therapy. The student would gain an understanding of how radioisotopes are produced, how to safely handle radioactive samples and how to evaluate radiation dose. Furthermore, the student would gain knowledge of how to prepare sterile materials for treatment. At the end of the program the student is expected to have gained broad insight in the field of radiation therapy, including hands on experience, and should be well poised to continue and deepen his or her knowledge in a specific area related to cancer treatment by medical isotopes.

Prerequisites/Student Eligibility:

Engineering, Chemistry or Physics major required. Students are expected to have taken basic chemistry, math and physics courses. Previous lab experience in chemistry or similar settings, either from courses or from individual research projects, is beneficial.

Project #1 – Development of Next Generation Spatial Transcriptomics

Description:

Each step-change in increasing resolution on RNA identification and quantification (i.e., transcriptome profiling) has been transformative to our understanding of cell biology. As the most recent technological revolution, the ability to map the spatial distribution of RNA molecules has revealed the importance of subcellular RNA localization, advanced our understanding of cell and developmental biology, and transformed the single cell transcriptomics field. However, despite the recent development, current RNA-mapping tools are limited to molecules with long specific sequence (mostly > 500 nt). Many other important RNA species (such as splicing isoforms, miRNA, and RNA editing), with much shorter specific sequence motif, remain inaccessible in situ. Our goal is to make this next resolution jump in profiling transcriptomics, i.e., RNA-mapping with high transcript- selectivity and single base-sensitivity.

Students’ Involvement and Expected Outcomes:

Students selected for this research project will study essential RNA mapping techniques in cell and tissue samples, collect single-molecule fluorescent in situ hybridization images using microscopies, and perform image-data processing.

Prerequisites:

Students with a molecular and cellular biology background are preferred. Experience in biochemistry and/or synthetic biology is a plus.

Project #1 – Applications for Laser Radiation for Medical Diagnostics

Description:

Dr. Venugopalan’s research focuses on laser-induced thermal, mechanical and radiative transport processes for application in medical diagnostics, therapeutics, biotechnology, and micro-electro-mechanical systems (MEMS). Dr. Venugopalan’s lab currently has three major research thrusts: 1) optical spectroscopic methods to non-invasively measure the optical and physiological properties of biological tissues for diagnosis and monitoring; 2) imaging of biological tissues using time-resolved optical interferometric methods; 3) development of pulsed laser microbeam methods for cellular manipulation and modification with applications for cellular microsurgery, microanalysis and MEMS.

Project #1 – Global Drought Monitoring and Prediction System

Project Description: 

Drought is a common climatic extreme that often spreads across a large spatial scale and spans over a long period of time. The economic damage of droughts across the United States on average is estimated as $6-8 billion annually. This indicates the importance of reliable drought monitoring, prediction and analysis tools in sustainable water resources management. The objective of this project is to validate and improve the currently available Global Drought Monitoring and Prediction System (GIDMaPS; http://drought.eng.uci.edu/).

Students’ Involvement:  

The project activates include: (a) validation and verification of GIDMaPS drought information over different regions (e.g., China); (b) analysis of trends and patterns of droughts across different spatial scales; (c) developing visualization portals to improve the way drought information is communicated globally. Students will develop skills in data processing, time series analysis, computer programing, Java scripting, and web design and development.

Prerequisites:  

All students familiar with basics of computer programming and web development are eligible. Relevant majors include: Computer Science; Civil and Environmental Engineering; Hydrology, Climatology.

Recommended Readings & Publications (optional): 
http://drought.eng.uci.edu/
http://amir.eng.uci.edu/
Hao Z., AghaKouchak A., 2013, A Nonparametric Multivariate Multi-Index Drought Monitoring Framework, Journal of Hydrometeorology, doi:10.1175/JHM-D-12-0160.1.

Project #1 – Low Impact Development of Small Scale Water Treatment System  

Description:  

Waterborne viral diseases are important burden to human society. A simple and low-cost method to inactivate viruses is needed to treat viral contamination in water for human uses. Nano-titanium dioxide (TiO2) coupled with solar radiation will be investigated to inactivate viruses in different types of water matrix using seeded coliphage. Light transmitting material will be explored as the support for TiO2 attachment.

Students’ Involvement and Expected Outcomes:

Students are required to conduct both literature research and field testing. Students will be trained for sampling technique and microbiological methods for viral assays. Probability based analyses will also be taught to students. Students involved in the project are expected to have a comprehensive understanding of the overall project, not only the tasks preformed by them. They are required to make a sound science-based presentation at the end of the research training.

Prerequisites:  

The ideal candidate should have fundamental knowledge of chemistry, biology, and preferable environmental microbiology. Statistical courses and math skills are also strongly desired. Students should be willing to participate in field-based research. Biological safety and lab safety training are required at the beginning of the research training.

Recommended Web sites and publications:

Water Res. 2011 Jan;45(2):535-44. doi: 10.1016/j.watres.2010.09.012. Epub 2010 Sep 19. Virus inactivation by silver doped titanium dioxide nanoparticles for drinking water treatment. Liga MV, Bryant EL, Colvin VL, Li Q.: http://www.ncbi.nlm.nih.gov/pubmed/20926111

Project #2 – Quantitative Risk Assessment of Marine Vibrio Related Human Illness

Description:  

Vibrios are gram negative, motile bacteria that can cause diseases in humans. They are commonly found in marine coastal ecosystems where their population changes with seawater temperature, increasing with warmer temperatures and algal blooms, while decreasing with cooler temperatures. Vibrio vulnificus and Vibrio parahaemolyticus were two of the most common vibrio infections reported in the United States between 1997 and 2006, responsible for the most vibrio related hospitalizations and deaths. With the ultimate goal of managing coastal resources and protecting human health, this research will develop a model to estimate the marine vibrio-related human illness. The objectives of this research are to: 1) Quantify the occurrence and distribution of vibrios along west coast of U.S. to generate database for vibrio concentrations and environmental parameters; 2) Perform quantitative microbial risk assessment (QMRA) to determine the human health risk of vibrio illness.

Students’ Involvement and Expected Outcomes:

Students are required to conduct both literature research and field sample collection and analysis for Vibrios. Students will be trained for sampling technique and microbiological methods for vibrio isolation and identification. Probability based analyses will also be taught to students. Students involved in the project are expected to have a comprehensive understanding of the overall project, not only the tasks preformed by them. They are required to make a sound science-based presentation at the end of the research training.

Prerequisites:  

The ideal candidate should have fundamental knowledge of chemistry, biology, and preferable environmental microbiology. Statistical courses and math skills are also strongly desired. Students should be willing to participate in field-based research. Biological safety and lab safety training are required at the beginning of the research training.

Recommended Web sites and publications:

Dickinson, G., K.Y. Lim, S. C. Jiang. 2013. Quantitative Microbial Risk Assessment of Pathogenic Vibrios in Marine Recreational Waters of Southern California. Applied and Environmental Microbiology. 79:294-392.

Project #1 – Development of Nanoparticle-Filled Magnetorheological Nano-composites for Semi-active Isolators

Description:  

Studies on magneto-mechanical responses for magnetorheological (MR) composites are of great interest to researchers and engineers in many science and engineering disciplines with civil engineering in particular since such smart materials can be employed for high power magnetostrictive actuation for anti-vibration applications, magnetoelastic sensor application in civil infrastructures monitoring, and on-demand damping control. The proposed research project aims to develop a novel type of MR nanocomposites filled with nanoparticles (such as carbon nanotubes and graphenes). Specifically, the project will focus on the composite fabrication process, three-dimensional tomography-based microstructural characterization, and dynamic magneto-mechanical testing of the MR three-phase nanocomposites. It is among the first attempt in the literature to combine the advantages of nanocomposites and magnetorheological materials to produce the novel smart nanocomposites in applications such as semi-active dampers and isolators for civil infrastructure systems. In addition, the project provides a unique opportunity for performing interdisciplinary research and cross-disciplinary education since it requires knowledge from engineering mechanics, materials science and civil structural engineering.

Students’ Involvement and Expected Outcomes:

Under the faculty mentor’s supervision, the selected student will conduct research on (1) materials fabrication process, (2) microstructural characterization using nano-CT system, and (3) dynamic magneto-mechanical analysis of the developed MR three-phase nanocomposites. It is expected that the undergraduate student gain hand-on research experience on nanocomposites development in the field of structural engineering for future graduate research programs.

Prerequisites:

Junior-standing undergraduate student majoring in engineering mechanics, civil engineering/structures, or materials science.

Recommended Web sites and publications: 

Li, R. and Sun, L.Z., 2011, “Dynamic mechanical behavior of magnetorheological nanocomposites filled with carbon nanotubes”, Applied Physics Letters, vol. 99, 131912-1-3.

Project #2 – Microstructure Quantification of Concrete and Cementitious Materials

Description:

Studies on mechanical performance of civil infrastructure materials are of great interest to researchers and engineers in science and engineering disciplines with civil and construction engineering in particular. The proposed research project aims to develop novel methodologies to quantifying microstructure and its evolution of concrete materials and/or ultra-high performance concrete materials using the state of the art three-dimensional X-ray computed tomography. The three dimensional microstructural damage characterization, pore network (porosity, pore size, and pore distribution) as well as the defects in the aggregates are specifically investigated. The project provides a unique opportunity for performing interdisciplinary research and cross-disciplinary education since it requires knowledge from engineering mechanics, materials science and civil structural engineering.

Students’ Involvement and Expected Outcomes:

Under the faculty mentor’s supervision, the selected student will conduct research on microstructural characterization using nano-CT system and modeling and simulation quantification using software package It is expected that the undergraduate student gain hand-on research experience on composite microstructure in the field of structural engineering for future graduate research programs.

Prerequisites:

Junior-standing undergraduate student majoring in engineering mechanics, civil engineering/structures, or materials science.

Recommended Web sites and publications:

sunlab.eng.uci.edu; Microstructural damage characterization of concrete under freeze-thaw action, International Journal of Damage Mechanics, October 2017, doi.org/10.1177/1056789517736573.

Project #3 – Development of Dielectric Elastomer Nano-composites as Stretchable Actuating Materials

Description: 

Dielectric elastomers are an emerging field of electro-active polymer materials attracting attention in the field of actuators in the last 10 years due to its flexibility and the ability to convert electrical energy to mechanical power and vice versa.We aim to address the fundamental issues associated with the development of dielectric elastomer nanocomposites (DENCs) filled with carbon nanotubes. The electromechanical responses of DENCs to applied electric fields will be investigated using computational materials science and multiscale materials modeling. We plan to demonstrate that a small amount of carbon nanotube fillers can effectively enhance the electromechanical performance of DENCs.

Students’ Involvement and Expected Outcomes:

Under the faculty mentor’s supervision, the selected student will conduct research on (1) microstructure characterization and (2) simulation of electro-mechanical coupling of DENCs. It is expected that the undergraduate student gain hand-on research experience on nanocomposites development in the field of mechanics and materials for future graduate research programs.

Prerequisites:

Junior-standing undergraduate student majoring in engineering mechanics and materials science.

Recommended Web sites and publications: 

Yu Wang and Lizhi Sun, Development of dielectric elastomer nano-composites as stretchable actuating materials, APPLIED PHYSICS LETTERS 111, 161904 (2017)

Project #1 – Design and Development of Autonomous Vehicles

Description:

The technology of autonomous vehicles (e.g. Google self-driving cars) offers the potential to reduce accidents, and save human lives. Autonomous vehicles relies on both sensor and wireless communication technologies to gather the information of surrounding environments. Based on the gathered information, the vehicle is able to understand its position, surround obstacles, and incoming traffics, thus enable the self-driving. In this project, the students will (1) implement a Vehicle to Vehicle (V2V) communication on the experimental vehicles, (2) implement a simple self-drive control algorithm based on the information from the V2V communications on the experimental vehicles.

Current Status of the Project: 

During 2016, a group of undergraduate students are building the following autonomous car as shown in the following figure.

This small vehicle is capable of decision-making at high-speeds of up to 65 km/h! In order to do this our autonomous car is utilizing the powerful Nvidia Jetson TK1 GPU to perform real-time computer vision calculations using CUDA. Our autonomous car is also utilizing state of the art sensors like the Hokuyo UST-10LX Scanning Laser Rangefinder (LIDAR) which will scan up to 30m around the car for obstacles.

Students’ Involvement and Expected Outcomes:

Students involved in this project will learn the basics of V2V communication in autonomous vehicles and the knowledge of control algorithms for autonomous vehicles. Technically, the students will gain the experience of electrical engineering and embedded software programming.

Prerequisites:

Engineering and basic knowledge of programming is required. Experience of embedded system design will be useful but not required.

Project #2 – Development of a Hybrid Energy Storage Prototype

Description:

Energy storage is a major component in state-of-the-art smart grids and vehicles that store and provide energy temporarily. However, the restricted energy stored in the system is the main bottleneck limiting the duration of the operation, e.g. driving range. Moreover, the energy characteristics of the storage (e.g. power density, energy density) influence the system operating parameters such as efficiency and reliability. Recently, hybrid energy storage has been introduced to leverage the energy characteristics of different energy storage (mainly electrical energy) in order to improve the operating parameters. The challenges involved in hybrid energy storage are designing the hardware components and managing the energy among the storage banks.

Current Status of the Project: 

We have built a battery-powered CPS Testbed in our lab. During this project, we will use this testbed for our experimental purposes during this project.

Students’ Involvement and Expected Outcomes:

In this project, the students will comprehend the basics of electrical energy storage and smart grid operation, in order to analyze their power request for different conditions. Moreover, they will learn energy management methodologies for monitoring and controlling hybrid energy systems. Meanwhile, the students are required to deploy a table-scale prototype and test bed for the hybrid energy storage and its corresponding management system.

Prerequisites:

Student with Electrical and Computer Engineering background and experience in programming and hardware design. Experience in system modeling and simulation (e.g. using MATLAB/Simulink) is a plus.

Project #3 – Data-Driven Analysis for DNA Synthesizing

Description:

Any living organism’s development, functionality, and growth are the result of executing genetic instructions embedded in DNA molecules inside the cells. Currently, one of the most controversial capabilities of the scientist is to artificially synthesize DNA molecules. The enabling technology for this advancement is made through a tight integration of cyber and physical components. This integration poses various vulnerabilities which can result in loss of intellectual property or distortion of integrity of the synthesized DNA molecules. In this project, students are going to evaluate such vulnerabilities through analysis of side-channel data. The main objectives of the research are (1) to explore various data that can be collected through the different sensors mounted around DNA synthesizer machine and develop algorithms to fuse them for feature extraction, (2) to design algorithms to estimate the state of the system for reliability.

Current Status of the Project:

A similar analysis has been conducted on a different application. Our work on 3D-printer received worldwide coverage. For details see https://www.youtube.com/watch?v=DlOHnp_gpGs&feature=youtu.be. In this work, we were able to recreate an object just from the sound the 3D-printer makes.

Students’ Involvement and Expected Outcomes:

Students will conduct research on digital signal processing, data fusion algorithms, and various algorithms to analyze data. They will understand the DNA synthesis mechanisms and develop skills in system forensics through data analysis.

Prerequisites:

Strong understanding of basic programming language (C, C++) and experience of developing a project with Matlab or Python is required. Basic knowledge of digital signal processing and data analysis is required. Knowledge of DNA synthesizers would enhance the experience, but is not required.

Project #4 – Data-Driven Approaches for Analyzing the Sensory Date from a 3D-Printer

Description:

Today’s fabrication systems have evolved with the capability to collect large amount of data from different sensors. As the demand for quality increases, it is imperative to leverage multiple data in monitoring and estimating (through various algorithms) the system performance for process/quality control. This project will focus on additive fabrication systems (3D-printing), which have gained wide acceptance in many sectors for fabrication. The main objectives of the research are (1) to explore various data that can be collected through the additive fabrication system (3D-printing) and develop algorithms to fuse them for feature extraction, (2) to design algorithms to estimate the performance of the system for process/quality control.

Current Status of the Project:

Our security-related work on 3D-printer received worldwide coverage. For details see https://www.youtube.com/watch?v=DIOHnp_gpGs&t=5s. In this work, we were able to recreate an object just from the sound the 3D-printer makes.

Students’ Involvement and Expected Outcomes:

Students will conduct research on digital signal processing, data fusion algorithms, and various data analysis algorithms. They will understand the framework of additive fabrication systems and develop skills in system forensics through data analysis.

Prerequisites:

Strong understanding of basic programming language (C, C++) and Matlab is required. Basic knowledge of digital signal processing and data analysis is required. Knowledge of additive fabrication systems would enhance the experience, but is not required.

Project #5 – DietMate – A Multimodal Diet Monitoring System

Description:

Healthcare has always been the key concern for everyone around the world. Significant amount of resources are invested to improve the medical treatments and mitigating the effect of various diseases. However, the preventive healthcare has always been neglected so far. Preventive healthcare allows for maintaining a balanced diet and overall a healthy life style by monitoring regular food intake and daily activities. Though there are lots of commercial devices available for monitoring daily activities but very less attention has been paid for monitoring the regular food intake. This project will focus on improving the techniques for diet monitoring. The main objectives of the research are- 1) Using piezoelectric sensors around the throat to gather data for different eating activities like swallowing, chewing etc. 2) Also use acoustic methods like throat microphone to obtain data. 3) Data generated from the above two modalities will be used to distinguish various eating activities as well as the type of food.

Current Status of the Project:

Currently our project is in the design phase where we plan to develop a smart collar with a bow tie. The bow tie will be equipped with all the necessary devices like piezoelectric sensor and microphone for data collection. Data collected from this modalities will be sent to a cloud where different data analysis algorithm will be implemented.

Final decision of the analysis will be displayed in the mobile interface. Though the design is under development however the expected prototype will look like given figure.

Students’ Involvement and Expected Outcomes:

Students will conduct research on embedded systems, digital signal processing, data fusion algorithms, and various data analysis algorithms. They will understand the framework of various health systems and develop skills in system forensics through data analysis.

Prerequisites/Student Eligibility:

Strong understanding of basic programming language (C, C++) and Matlab is required. Basic knowledge of digital signal processing, hardware design and data analysis is required. Knowledge of coding in android platform is a plus, but is not required.

Please see the following posters for further information about Prof. Al Faruque’s work:

Project #1 – Vehicle Telemetry Date Analysis System and Test-bed for Driver Training and Object Detection

Description: 

This project is a culmination of several undergraduate projects over the past couple of years consisting of both software and hardware. This past summer, UCInspire students worked with our undergraduate juniors and seniors involved in the FSAE program by building a proof-of-concept data analysis software to display custom vehicle telemetry data. The drivers and coaches were able to use this information for driver training and improvement. Students also began prototyping new uses of off-the-shelf hardware sensors for augmenting existing data with object detection for display in the software, which no current commercial product implements out-of-the-box.

This coming year, we will be building upon that knowledge in different fields:

• improving the user interface and algorithms to display data more smoothly and focus on usability by non-experts
• overlaying data onto synchronized video on-the-fly for increased marketability
• streaming data and video wireless in real-time through customizing off-the-shelf hardware and writing the translation software to interpret the wireless data on a local device
• creating a hardware based test-bed that will emulate a vehicle for easier testing of prototype software and hardware sensors

Project #1 – Emerging Technologies in Precision Medicine

Description:

Malaria is one of the life-threatening and wide-spread diseases in the world transmitted by the bite of infected mosquitoes. According to WHO, 219 million cases and 435000 deaths due to malaria were reported all over the world in 2017. In order to solve this problem, we are working on a novel, inexpensive and reusable microfluidic platform for rapid, label-free detection of malarial parasites using impedance measurements of Red Blood Cells. For this project, the student will use a silver Nano-particle ink and an inkjet printer to fabricate the electronic apparatus on flexible PET substrate. Further, the student will use an insulating layer between the electrodes and the PDMS microchip for contactless impedance measurements, thus eliminating electrode fouling and making the electronic apparatus reusable. For validation experiments, the student will use polystyrene beads (10micro meter) as test particles to mode Red Blood Cells presence.

Project #1 – R2AD: Randomization and Reconstructor-based Adversarial Defense on Deep Neural Network

Description:

Machine learning (ML) has been widely adopted in a plethora of applications ranging from simple series forecasting to computer security and autonomous systems. Despite the robustness by the  ML  algorithms  against  random  noise,  it  has been  shown  that  inclusion  of  specially  crafted  perturbations  to the  input  data  termed  as adversarial  samples can  lead  to  significant degradation in the ML performance. Existing defenses to   mitigate   or   minimize   the   impact   of   adversarial   samples including  adversarial  training  or  randomization  are  confined to  specific  categories  of  adversaries,  compute-intensive  and/or often  lead  to  reduce  performance  even  without  adversaries.  To overcome the shortcomings of the existing works on adversarial defense,  we  propose  a  two-stage  defense  technique  (R2AD). To  minimize  the  exploitation  of  the  network  by  the  attacker,we  first  include  a  random  nullification  (RNF)  layer.  The   NF nullifies/removes  some  of  the  features  from  the  input  randomly to reduce the impact of adversarial noise. However, the removal of  input  features  through  RNF  leads  to  a  reduction  in  the performance  of  the  ML.  As  an  antidote,  we  equip  the  network with  a  Reconstructor.  The  Reconstructor  primarily  contributes to  reconstructing  the  input  data  by  utilizing  an  autoencoder network  but  based  on  the  distribution  of  the  normal  samples,thereby improving the performance, and also being robust to the adversarial  noise.  We  evaluated  the  performance  of  proposed multi-stage R2AD on  the  MNIST  dataset  against  multiple  ad-versarial  attacks  including  FGSM,  JSMA,  BIM,  Deepfool,  andCW attacks. Our findings report improvements as high as 80%in  the  performance  compared  to  the  existing  defenses  such  as adversarial  training  and  randomization-based  defense.

Prerequisites: 

Machine Learning, Computer Vision (Image Processing + ), Basic knowledge of chip fabrication

Project #2 – IPF Autonomous Vehicle Example

Description:

We will create a framework for autonomous vehicles, enabling a deep analysis. The framework consists of metrics for sensors, reinforcement learning based actuation control, and optimization techniques both for offline (DSE) and online. The framework not just considers the current status of the vehicle but predicts future events based on the past data.

Prerequisites: 

Machine Learning Techniques (TensorFlow/Keras/CNN/YOLO), Low-level programming (scheduling, daemonize, optimization), Basic understanding of Computer Architecture

Project #3 – Mobile Apps Development

Description:  

Mobile programming is becoming a major component in today’s training of Computer and Communication Engineers.

Students’ Involvement and Expected Outcomes:

Students selected for this research project will study the Java language and the Eclipse development system. They will develop apps for Android phones/tablets that can make use of the available sensors in a typical platform and perform useful tasks.

Prerequisites:  

Students with Computer Engineering background and experience in computer programming. Experience in using JAVA and Eclipse programming environment is a plus.

Project #1 – AsterixDB

Description:

Apache AsterixDB [9] is a Big Data Management System (BDMS) with a feature set chosen to target use cases such as web data warehousing and social media data analysis. The system has been co-developed at UC Irvine and UC Riverside and activities continue at both as well as in the Apache open source community. Its notable features include:

1) A NoSQL-style data model (ADM) based on extending JSON with object database concepts;

2) A declarative query language (SQL++) that supports a broad range of queries against multiple semi-structured datasets;

3) A rule-based, data-partition-aware query optimizer for parallel queries (Algebricks);

4) An efficient dataflow execution engine (Hyracks) for partitioned-parallel execution of query plans;

5) Partitioned and LSM-based native storage and indexing for large datasets;

6) Support for querying and indexing of external data (e.g., data in HDFS) as well as natively stored data;

7) Rich data type support, including numeric, textual, temporal, and simple spatial data;

8) Secondary indexing through B+ trees, R-trees, and several variants of inverted keyword indexes;

9) Basic NoSQL-like transactional capabilities similar to those of popular NoSQL stores.

Students’ Involvement and Expected Outcomes:

Students will assist UCI PhD students with various AsterixDB-related research projects.

Prerequisites:

Students need a Computer Science/Engineering background and experience in computer programming in Java and/or Python. Some experience with SQL-based database systems is also required. Experience with database system internals, NoSQL databases such as MongoDB or Apache Cassandra, and/or data analytics platforms such as Apache Spark would be a plus

Project #1 – SCENORITA – Generating Diverse, Fully-Mutable, Test Scenarios for Autonomous Vehicle Planning

Description:

Autonomous Vehicles (AVs) leverage advanced sensing and networking technologies (e.g., camera, LiDAR, RADAR, GPS, DSRC, 5G, etc.) to enable safe and efficient driving without human drivers. Although still in its infancy, AV technology is becoming increasingly common and could radically transform our transportation system and by extension, our economy and society. As a result, there is tremendous global enthusiasm for research, development, and deployment of AVs, e.g., self-driving taxis and trucks from Waymo and Baidu. The current practice for testing AVs uses virtual tests—where AVs are tested in software simulations—since they offer a more efficient and safer alternative compared to field operational tests. Specifically, search-based approaches are used to find particularly critical situations. These approaches provide an opportunity to automatically generate tests; however, systematically creating valid and effective tests for AV software remains a major challenge. To address this challenge, we introduce SCENORITA, a test generation approach for AVs that uses evolutionary algorithms with (1) a novel gene representation that allows obstacles to be fully mutable, hence, resulting in more reported violations, (2) 5 test oracles to determine both safety and motion sickness-inducing violations, and (3) a novel technique to identify and eliminate duplicate tests. Our extensive evaluation shows that SCENORITA can produce effective driving scenarios that expose an ego car to safety critical situations. SCENORITA generated tests that resulted in a total of 1,026 unique violations, increasing the number of reported violations by 23.47% and 24.21% compared to random test generation and state-of-the-art partially-mutable test generation, respectively.

Project #2 – Doppelgänger Test Generation for Revealing Bugs in Autonomous Driving Software

Description:

Vehicles controlled by autonomous driving software (ADS) are expected to bring many social and economic benefits, but at the current stage not being broadly used due to concerns with regard to their safety. Virtual tests, where autonomous vehicles are tested in software simulation, are common practices because they are more efficient and safer compared to field operational tests. Specifically, search-based approaches are used to find particularly critical situations. These approaches provide an opportunity to automatically generate tests; however, systematically producing bug-revealing tests for ADS remains a major challenge. To address this challenge, we introduce DoppelTest, a test generation approach for ADSes that utilizes a genetic algorithm to discover bug-revealing violations through generating scenarios with multiple autonomous vehicles that account for traffic control (e.g., traffic signals and stop signs). Our extensive evaluation shows that DoppelTest can efficiently discover 123 bug-revealing violations for a production-grade ADS (Baidu Apollo) which we then classify into 8 unique bug categories.

Project #3 – Path-Sensitive Analysis of Message-Controlled Communication for Android Apps

Description:

To support quality development of Android apps, a variety of techniques have been produced for analyzing the exchange of messages, i.e., Intents, among Android components. Intents and their payloads can cause a variety of operations to be performed, and can be filled with malicious data, demonstrating that Intents can serve as attack vectors of an insecure app. Intents may further guard or control execution of different program paths, which may contain vulnerable, faulty, or energy-inefficient code. While different techniques have focused on determining possible Intents in an app, none have focused on analyzing Intents per program path, i.e., path-sensitive Intent analysis. Analyzing a program per path allows the determination of the attributes of Intents needed to control execution of a program from its message-based inter-component interface. Unfortunately, analyzing apps in a path-sensitive manner faces scalability issues. To address these challenges, we introduce a novel, scalable framework called PHENOMENON (PatH-sEnsitive aNalysis Of MEssage-coNtrOlled communication for aNdroid apps). We evaluate the accuracy of PHENOMENON’s path-sensitive analysis on five apps with 4KSLOC–460KSLOC, over a total of 4,100 program paths, achieving an accuracy of over 96% for each app. To evaluate PHENOMENON’s efficiency, we assess it on 100 randomly selected apps, demonstrating an average runtime of 30 seconds, with no app taking more than 180 seconds to analyze.

Project #4 – Native Bomb: Automatic Exploit Generation of Native Vulnerabilities in Android Apps

Description:

Third-party libraries are an integral part of mobile apps. Android developers opt for third-party libraries due to their convenience and re-usability, since utilizing them saves time and effort and allows developers to avoid re-implementing functionality. Furthermore, native libraries have become more prevalent in recent Android applications (“apps”), especially social networking and gaming apps. These two categories of apps—which ranked among the top categories on Google Play, the official Android app market—require special tasks such as 3D rendering, and audio/video encoding/decoding. These tasks tend to be resource-intensive and are, thus, often handled by native libraries to improve runtime performance. 

Despite the convenience and benefits that mobile developers obtain from third-party libraries, they can expose end-users to a wide range of security attacks. The ubiquity of third-party libraries in Android apps increases the attack surface since host apps expose vulnerabilities propagated from these libraries. However, these previous work does not consider native libraries used by Android apps. We argue that security implications in native libraries are even more critical for three main reasons since developers add native libraries but do not update them, they are susceptible to memory vulnerabilities (e.g., buffer overflow attacks) that are very difficult to exploit with managed code of Android apps, i.e., Dalvik code, and native libraries are currently used pervasively in top mobile apps.

To study this neglected attack vector, we obtained 7,328 versions of the top 200 free apps from Google Play, collected between Sept. 2013 and Feb. 2020. On average, there are 40 versions per app with up to 166 versions for com.instagram.android. From these apps, we identified 67,389 native libraries in total with an average of 11 libraries per app, with up to 141 libraries in an app. From this repository of apps and libraries, we identified 57 apps with 1,788 vulnerable versions with known CVEs between Sept. 2013 and Feb. 2020. 26 of these apps remain vulnerable. We further find that these apps have a long period of outdatedness, on average, of 1,202.48 ± 63.42 days.

Using the aforementioned repository of apps with vulnerable native libraries, we propose to produce a novel technique called NativeBomb that automatically generates exploits for vulnerabilities in native code of Android apps. To that end, we aim to build a scalable cross-language analysis that can produce inputs capable of reaching vulnerable native code from the Dalvik managed code of an Android app. 

Project #5 – Microservice Architecture Recovery and Pattern Detection of Orchestrated System

Description:

Microservice architecture has become widely-used in industry, with tech giants like Amazon, Twitter, and LinkedIn leveraging microservices to evolve their web-scale applications. Microservice architecture brings benefits such as scalability and technological heterogeneity, although at a cost of complexity. A number of microservice patterns have been proposed to address this cost. Detecting such patterns can aid with the architectural maintenance of microservice systems. To detect such patterns in existing systems, techniques are needed to recover microservice architectures from such systems. However, there have been few attempts at recoveries of microservice systems. To address these concerns, this paper introduces (1) a recovery technique for orchestrated, microservice systems, which has been applied to three benchmark systems; and (2) a microservice pattern detection technique for five patterns. We recover and detect microservice patterns on three microservice systems. From our recoveries, we identified between 6% to 11% of the systems have decayed in terms of components, while 25% to 44% of the systems have decayed in terms of connectors. For pattern detection, our techniques found 4 of the 5 patterns in all three benchmark systems.

Project #6 – A Comprehensive Study of Designing Architectural Modules in Modern Java Systems and Tool Support for Automated Java Module Creation

Description:

With the release of Java 9 and its novel Java Platform Module System (JPMS) in 2017, software architects and engineers have been empowered with the ability to specify software architectural modules in the Java language itself, including the ability to control the required and provided interfaces of a module at different granularities (e.g., from single classes or interfaces to entire packages) and for compile-time and run-time access. These features enable an engineer to allow Java systems to have stronger encapsulation, improved memory utilization, and enhanced security. However, designing such modules remains challenging, preventing wide adoption of Java modules. To address these challenges in designing architectural modules that strongly connect with code, we aim to study the design of architectural Java modules in existing software systems (e.g., JUnit, Microsoft Azure SDK, and JDK implementations) to determine how they currently modularize their systems and how this modularity can be improved. Based on lessons learned from this study, and pre-existing knowledge about quality design of architectural modules, we aim to produce tools to automatically migrate or improve the modularization of existing Java systems. Through our preliminary experiments, we have found that architecture recovery techniques (e.g., ACDC and ARC) have recovery criteria that are extensively used by existing Java modules of widely-used systems (e.g., JUnit or Microsoft Azure SDK). Thus, we hypothesize that these recovery tools can aid in the modularization of recovery techniques that are at least as well-modularized as that of software architects or developers of these systems. 

Project #7 – Automated Configuration Testing for Autonomous Vehicle Software

Description:

An autonomous vehicle (AV) system is a kind of highly configurable system that consists of numerous options. By tuning the values of different configuration

options of such a system, we can determine the performance and assess the functionalities of AVs. Automated testing is an important research direction in the AV domain for enhancing the security and safety of self-driving cars. Currently, AV systems  are  widely  tested  in  the  virtual  environment  to simulate real-world scenarios. In our research, we seek to find an automated way to test AV systems in different configuration settings. We are currently evaluating our experiment by setting multiple objectives, including the number of violations triggered by the scenario, code coverage, and execution time. We are assessing different optimization algorithms, including a genetic   algorithm and deep reinforcement learning, to tune the options and search for the optimal configuration that meets the three objectives. We utilize various driving scenarios consisting of the planning route of the ego car and obstacles, including pedestrians and vehicles, to test the configurations. The search space of configurations in an AV system is very large (i.e., contains thousands of options). To prune this space, we apply static analysis to find the dependencies among the options at the code level, which we use to determine the number of combined configuration options that need to be tested in tandem. We are evaluating our approach using two prominent AV systems, Apollo and Autoware. Additionally, our work has resulted in contributions to the Apollo open-source project.

Project #8 – Automatically Reproducing Scenarios from AV Bug Reports

Description:

Testing is an essential aspect of automated driving system (ADS) development that ensures the vehicles driven by self-driving software are safe. While testing an ADS on physical streets and highways fails to  capture  many  rare  events,  existing  simulation-based  testing  methods  could  discover unexpected corner-case traffic violations of ADS under complex driving scenarios that require sophisticated  awareness  of  the  surroundings.  As  simulation-based  scenarios  become  more 

popular, users are increasingly making use of screen recordings as a means to report bugs (e.g., violations)  to  ADS  developers.  Meanwhile,  debugging  ADS  based  on  a crowdsourced feedback mechanism requires developers to manually inspect collected  users’ video recordings to reconstruct  the  same  driving  scenarios.  However,  it  is  still  tedious  and  time-consuming  for developers  to  reproduce  the  bug  due  to  the  length  and  complex  actions  (e.g.,  obstacle maneuvers) that happen within a scenario. 

To  overcome  these  issues,  our  project  proposes  an  approach  to  automatically  reproduce simulation-based scenarios from bug reports by leveraging both visual and textual information. Given the importance of visual information in interpreting complex scenarios, our method could automatically analyze video recording(s) from a bug report to extract traffic actors and their states in consecutive frames. Those traffic actors are then re-simulated along with an ego car driven by an ADS under the same scenario to trigger the violations. To enhance the accuracy of the reconstruction process: (1) the quality of each produced scenario will be progressively compared with the one in the bug report video by estimating their visual similarity and (2) the precision of discovered violations will be verified against the original textual description.

Project #9 – SAHARA: Software Analysis for Hardware Acceleration using Reconfigurable Architectures

Description:

Tackling the most challenging problems of human knowledge (e.g., studying the human brain, climate change, and the structure and evolution of the universe and of space-time itself), requires exceptional advances in how these problems are described to, and solved by, software and computers. Software enables (1) new discoveries from mining billions of objects observed by astronomical surveys, which are routinely measured using terabytes or petabytes, with raw data volumes already measured using exabytes and (2) new physical insights from the connection of these enormous data volumes to petaFLOPS and exaFLOPS of computing power to simulate them from multiple physical theories and scenarios. However, even with the current convergence of High Performance Computing (HPC) and data-driven machine learning workloads, making use of HPC facilities to benefit these scientific applications is a challenging task that is only increasing in difficulty.

We are currently seeing a new era in the development of general processor architectures with the appearance of new instruction set architectures (ISAs) in the HPC field, such as A64FX ARM and RISC- V, besides the co-existence of established architectures such as AMD64 and POWER. These new ISAs in HPC add to the ever increasing heterogeneity of accelerators (e.g., classic GPUs, Tensor Processing Units, FPGAs, and Neuromorphic and Quantum PUs). For scientific applications to take advantage of the full power of HPC, they must capture not only the natural complexity of the problem they target—which mapping into software is already in many cases non-trivial—but also all the added complexities related to optimizing resource utilization, sometimes down to the choice of hardware.

Even with all this complexity, in most HPC environments, practicing scientists so far prefer to rely on explicit control of the platform and accelerators, and to explicitly drive code parallelism. Even with all the investment and efforts in the past decades in tools and software development patterns for parallel computing, OpenMP, MPI, OpenCL, and CUDA are still the preferred technologies used by domain scientists when developing high performance applications.

Explicitly analysing and (re-)writting the code to take advantage of parallelism using these technologies has three fundamental challenges in the upcoming HPC era. First, dealing with code scalability with respect to the number of nodes and cores, which due to limits on the transitor sizes and operating frequencies, is becoming more difficult as core counts increase. Second, there is an ever increasing heterogeneity of the target hardware as described before, which so far has been mostly constrained to AMD64 CPUs and GPUs, increasing the complexity of the memory hierarchy. Third, the complexity of writing code to leverage the power of modern reconfigurable accelerators is very high.

Project #10 – Modularization of C++ Applications Based on C++ 20 Modules

Description:

As one of the most popular programming languages, C++ is characterized by its unique header-file mechanism that provides an effective way to access the interface of a library. However, this header-based mechanism also has its weaknesses. For instance, compilers have to perform redundant work which leads to poor compiling performance, developers should write their code carefully to avoid macro collisions, declarations and implementations are separated into multiple files which increase the complexity of the project, etc. To mitigate these challenges, the C++ standards committee proposed the modules feature of C++20 in 2020, which is introduced as an C++20 modules provide a better way to encapsulate codes and address most deficiencies of header files. However, since the module feature is quite new, its advantages and potential challenges are not well-understood. On the other hand, most existing C++ applications are still based on header files and the include model, and there are not enough instructions on how to modularize a header-based app into a module-based app. To bridge these gaps, the paper discusses the influence of C++20 modules and proposes H2M, a new approach for conversion of a header-based C++ app to a module-based C++ app with better compiling performance. H2M starts by determining candidate source files for modularization. Next, it bundles up similar candidate source files and identifies appropriate dependencies. Finally, H2M generates the corresponding module-based app of the given header-based app. Our empirical studies verify the effectiveness of C++ 20 modules in improving compiling performance and the feasibility of code migration. Besides, the empirical studies on several header-based C++ applications demonstrate the effectiveness of H2M.

Project #1 – Natural Language Processing

Description:

Research projects in Professor Harris’ group is related to testing of hardware and software systems. His field of interest includes validation of hardware systems to ensure that the behavior of the system matches the intentions of the designer. He also investigates the application of testing for computer security. Natural Language Processing (NLP) is a prominent theme in Professor Harris’ work in both security and verification. NLP techniques are used to extract information from hardware specifications, and NLP techniques are used to identify social engineering attacks in a dialog between two speakers.

Project #1 – Resilient Distributed Computing

Description:

The objective of the project is the development of an advanced technology to reliably distribute computation tasks in autonomous copter systems. The ability to observe and analyze the surrounding environment is the key to autonomy. However, advanced data processing is an intense and resource consuming task, especially when contextualized in copter platforms, where weight constraints may limit the availability of vital resources such as energy and processing power. Running complex algorithms on-board may result in limited analysis capabilities, long data capture-to-control time, and reduced mission lifetime due to battery depletion. Traditional approaches in mobile computing differentiate the capabilities of the devices, with the extreme points being energy and processing constrained mobile devices and powerful, but static and shared, cloud data centers.

Student’s Involvement and Expected Outcomes:

The student will be involved in the
experimentation of the platform and software developed during the project. Experiments will
include the deployment and supervision of a group of copters autonomously operating in a
predefined outdoor space. The student will assist data collection and processing to extract
performance metrics. Depending on his/her background, the student will be assigned
development tasks in autonomous navigation programming or data analysis components of
the project.

Student Background:

Python, experience in quadcopter/robotics or machine vision is a plus.

Recommended Readings:

The student should familiarize with tools used in experiments:
– Dronekit: high level API used to control the UAVs
– SITL: intermediate layer used in simulations
– FlightControllers and mavproxy: layer in between Dronekit and the actual hardwareAdditional

Project #1 – Big Data Visualization Using Cloudberry 

Description: 

Cloudberry is an open resource research project on supporting visualization of big data. We intend to develop novel techniques to allow large amounts of data to be efficiently visualized by users. We use Cloudberry to develop the end-to-end application called TwitterMap to support interactive analytics and visualization on more than one billion social media tweets, which are semantically rich with temporal, spatial, and textual attributes. The system is currently used by researchers from different domains and multiple universities for their own studies.

Students’ Involvement and Expected Outcomes:

Under the faculty mentor and graduate students’ supervision, the selected student will design and implement advanced features on the frontend, and invent novel techniques on the backend to support scalable visualization.

Prerequisites:

Junior-standing undergraduate student majoring in Computer Science. Familiar with frontend techniques including HTML, CSS, and JavaScript.

Recommended Web sites and publications:

• http://cloudberry.ics.uci.edu/
• http://cloudberry.ics.uci.edu/apps/twittermap/

Project #1 – Efficient Magnetic Design for Power Electronics Converters

Description:

Magnetics is a crucial component for high frequency power conversion. Proper configuration and design will enhance the efficiency and reduces the noise of the converters.
In this project students will conduct a survey about magnetics including transformers and inductors, perform simulation of selected group configurations, build experimental prototypes.

Project #1 – AnatomyNet: Deep Learning for Fast and Fully Automated Whole-volume Segmentation of Head and Neck Anatomy

Description:

Radiation therapy (RT) is a common treatment option for head and neck (HaN) cancer. An important step involved in RT planning is the delineation of organs-at-risks (OARs) based on HaN computed tomography (CT). However, manually delineating OARs is time-consuming as each slice of CT images needs to be individually examined and a typical CT consists of hundreds of slices.Automating OARs segmentation has the benefit of both reducing the time and improving the quality of RT planning. Existing anatomy auto-segmentation algorithms use primarily atlas-based methods, which require sophisticated atlas creation and cannot adequately account for anatomy variations among patients. In this work, we propose an end-to-end, atlas-free 3D convolutional deep learning framework for fast and fully automated whole-volume HaN anatomy segmentation.

Project #1 – Atomic-scale Structure-property Relationships of Advanced Functional Materials

Description:

Pan’s research interests center on understanding the atomic-scale structure-property relationships of advanced functional materials, including oxide electronics, nanostructured ferroelectrics and multiferroics, and catalysts. He is recognized internationally for his work in materials physics and electron microscopy that have led to the discovery of new properties and novel functionalities in technologically important materials. His pioneering contributions include the development of methods to quantitatively map the electrical polarization in ferroelectrics at atomic resolution, and methods to uncover the effects of boundary conditions on ferroelectricity, including polarization mapping, first observation of ferroelectric vortices, and dynamic behaviors of ferroelectric domains during electrical switching under applied electric field in TEM.

Project #1 – Dielectrophoretic Electropolymerization for Particle Entrapment  

Description:  

Dielectrophoretic Electropolymerization is a novel technology developed in the BioMEMs research Lab at the University of California, Irvine. This technology is used to attract and entrap microparticles on conductive surfaces to create high specific surface area electrodes. Polymeric microbeads, silicon particles, and biological cells (such as yeast cells) can be used to pattern the surface of the electrodes.

Students’ Involvement and Expected Outcomes:

Students will perform experiments in sequential dielectrophoresis and electropolymerization in pyrrole solution with polystyrene microbeads and yeast cells and study parametric conditions (voltage, frequency, flow rate) for optimized patterning of microelectrodes.

Prerequisites:  Consent of Research Mentor.

Project #1 – Clean Water Generation in the Ambient Environment using Selective Emitters

Description:

Radiative thermal management in the ambient environment holds great potential for solving global energy and water problems. Substantial surface cooling or heating without electricity is possible by controlling solar absorption and infrared emission phenomena with selective emitters. The project will focus on developing selective emitters based on engineered materials and incorporating selective emitters for clean water generators such as ambient water condensation and solar-powered membrane distillation systems.

Students’ Involvement and Expected Outcomes:

Students selected for this project will investigate optimal selective emitter designs for radiative thermal control, perform thermal characterization using an infrared camera and thermocouples, and test performance of water generation systems in the ambient environment. Under the faculty mentor’s supervision, the students are expected to learn about thermal radiation and water generation analysis techniques, gain hands-on experience with novel materials and advanced thermal metrology systems, and interact with other students and researchers in the Nano Thermal Energy Research (NTER) Lab.

Prerequisites:

Excellent academic record in mechanical, chemical, environmental, civil, or materials science engineering.

Recommended Web sites and publications:

nter.eng.uci.edu
A. Krishna, and J. Lee, “Morphology-Driven Emissivity of Microscale Tree-like Structures for Radiative Thermal Management,” Nanoscale and Microscale Thermophysical Engineering, vol. 22, no. 2, 124–136, (2018).
P.D. Dongare et al., “Nanophotonics-enabled solar membrane distillation for off-grid water purification” Proc. Natl. Acad. Sci. USA, 114, pp. 6936-6941, (2017).

Project #2 – Bio-Inspired Nanostructures for Thermoregulation and Thermal Camouflage System

Description:

This project will target developing novel thermoregulation and thermal camouflage systems by replicating unique surface designs in nature. For instance, some desert ants are found out to stay cool under strong sunlight due to their triangular hair structures that are highly emissive in the mid-infrared. On the other hand, polar bears are known to keep their bodies warm in the cold due to a white hairy fur that minimizes radiative heat losses. Chameleons are known for dynamic skin color change, and the mechanism is recently attributed to the effect of photonic crystals. Inspired by these structural thermal control mechanisms, the project seeks to engineer the topography of surfaces and demonstrate nanostructure-based thermoregulation and thermal camouflage systems.

Students’ Involvement and Expected Outcomes:

Students selected for this project will investigate structural coloration and thermal control mechanisms in nature, design bio-inspired nanostructures for optimal thermoregulation and thermal camouflage, and perform thermal characterization of various surface materials using an infrared camera and thermocouples. Under the faculty mentor’s supervision, the students are expected to learn about electromagnetic computation methods, gain hands-on experience with nanostructured surface materials and advanced thermal metrology systems, and interact with other students and researchers in the Nano Thermal Energy Research (NTER) Lab.

Prerequisites:

Excellent academic record in mechanical, chemical, environmental, civil, or materials science engineering.

Recommended Web sites and publications:

nter.eng.uci.edu
A. Krishna, and J. Lee, “Morphology-Driven Emissivity of Microscale Tree-like Structures for Radiative Thermal Management,” Nanoscale and Microscale Thermophysical Engineering, vol. 22, no. 2, 124–136, (2018).
N. N. Shi et al., “Keeping Cool: Enhanced Optical Reflection and Radiative Heat Dissipation in Saharan Silver Ants.” Science, 349, 298–301, (2015).

Project #1 – Security and Privacy of Graph-based Machine Learning

Description:

Graph-based Machine Learning (GML) has demonstrated its impacts in real-world applications that handle graph-structured data, like social networks. In this project, the students are expected to apply GML for cyber-security applications, like intrusion detection. The students will develop new methods and frameworks that achieve good efficiency and effectiveness on the benchmark datasets. In addition the security perspectives, the students can also explore the privacy perspectives of GML, e.g., how to avoid information leakage from GML learning and inference by adjusting the methods under differential privacy.

Students’ Involvement and Expected Outcomes:

Study – understanding GML models like Graph Convolutional Network (GCN).
Implementation – write machine-learning code to build new frameworks and experiment with them.

Prerequisites:

Consent of Research Mentor. Prior experiences with machine-learning environment, like PyTorch.

Recommended Web sites and publications:

Euler: Detecting Network Lateral Movement via Scalable Temporal Graph Link Prediction, NDSS’22
shadewatcher: recommendation-guided cyber threat analysis using system audit record, IEEE S&P’22
Communication Efficient Triangle Counting under Local Differential Privacy, Usenix Security’22

Project #2 – Automated Vulnerability Discovery on Network Software

Description:

Network software like Domain Name System (DNS) stubs and resolvers are essential to power people’s daily Internet activities. However, security bugs (or vulnerabilities) were frequently identified from them, which could lead to hacks that harm users. This project aims to develop new methods to automatically discover the vulnerabilities by analyzing the software code or binary. The methods are preferably to be built on top of software fuzzing.

Students’ Involvement and Expected Outcomes:

Study – understand network protocols work and learn how to use the fuzzing tools.

Implementation – customize the existing fuzzing methods to network software analysis, test the software and analyze the root causes of the vulnerabilities.

Prerequisites:

Consent of Research Mentor. Prior experiences with software analysis and computer networks.

Recommended Websites and Publications:

https://github.com/google/AFL

https://github.com/aflnet/aflnet

Project #1 – Machine learning in Multi-scale Wearable Device Datas

Description:

Wearable devices are becoming increasingly popular as health status
monitors. Data is collected on a variety of factors such as heart rate, blood pressure,
stress levels, etc. However, different sampling frequencies of various factors lead to
multiscale data structures which signifies the difficulty of data analysis and
downstreamed tasks, e.g. predicting real-time stress level given multiscale heart rate
and blood pressure data. The goal of this research project is to predict the health
outcome by leveraging the nature of multi-resolution data.

Students’ Involvement and Expected Outcomes:

Students working on this project would
have the opportunity to be involved in a range of activities related to statistical research.
The students will learn about wearable health monitors data and gain an understanding of
how data is produced, how to clean and integrate data from different sources, and how to
analyze the data. Furthermore, the student would gain knowledge of how to handle multi-
scale data, which plays an important role in medical, finance, meteorology, physics, and
industrial fields like process monitoring, fault diagnosis. Statistical learning analyses will be
taught to students. Students involved in the project are expected to have a comprehensive
understanding of the overall project, not only the tasks performed by them. They are
required to make a sound science-based presentation at the end of the research training.

Prerequisites:

A strong understanding of the programming language R or Python is
required. Basic knowledge of statistics and data analysis is required. Prior experiences with
machine-learning environments, like TensorFlow and numpy.

Recommended Web sites and publications:

Sun, C., Hong, S., Song, M., & Li, H. (2020).
A review of deep learning methods for irregularly sampled medical time series data. arXiv
preprint arXiv:2010.12493.

Ferreira, M. A., Higdon, D. M., Lee, H. K., & West, M. (2006). Multi-scale and hidden
resolution time series models. Bayesian Analysis, 1(4), 947-967.

Project #2 – Active Learning through Human and Machine Interactions

Description:

Another area we are working on is the active learning which arises from the practical problems involving large-scale labeling. Active learning is a modern smart way to train machines to learn from minimum human feedbacks so that the cluster labeling can be done efficiently and accurately without extensive human labor. Specifically, we solve active clustering problem with sequentially queried pairwise constraints between samples. In contrast to traditional semi-supervised clustering approaches which incorporate only constraints within part of the data, we propose a novel query augmented method to generate implicit structure-level constraints onto the entire dataset, which is more effective in learning a metric that favors the clustering task. The major advantages of the proposed augmenting procedure are that it captures richer constraint information while utilizing the history training results to refine the feature space. In addition, we propose a new active strategy to query the most informative pairs adaptively based on the expected entropy change. This leads to a more efficient clustering process without extra labeling cost.

Students’ Involvement and Expected Outcomes: 

Students working on this project would have the opportunity to be involved in a range of activities related to statistical research. The students will learn about active learning for high-dimensional data. Furthermore, the student would gain knowledge of how to handle large-scale supervised learning or semi-supervised learning, which plays an important role in medical, finance, meteorology, physics, and industrial fields like process monitoring, fault diagnosis. Students involved in the project are expected to have a comprehensive understanding of the overall project, not only the tasks performed by them. They are required to make a sound science-based presentation at the end of the research training.

Prerequisites:  

A strong understanding of the programming language R or Python is required. Basic knowledge of statistics and data analysis is required. Prior experiences with machine-learning environments, like TensorFlow and numpy.

References:

Deng, Y.*, Yuan, Y.*, Fu, H. and Qu, A. “Query-augmented active metric learning.” Journal of the American Statistical Association, to appear.

Project #1 – Fetal and Maternal Monitoring

Description:

Our group develops abdominal patches worn by pregnant subjects for home-based monitoring of the mom and the fetus. The system includes an unobtrusive patch with non-contact electrodes for collecting abdominal ECG, a microcontroller for signal processing and a Bluetooth Low Energy module for wireless communication with a smart device. In this big project, there are several sub-projects that UCInspire students can do. For example:

1. Design and implementation of the flexible patch using an origami-inspired circuitry for signal acquisition
2. Edge-based processing and analysis of the fetal and maternal ECG signals
3. App design and cloud computing for fetal ECG extraction and processing

Project #2 – Acquisition and analysis of electroencephalogram (EEG)

Description:

This project involves the implementation of an EEG helmet system with embedded machine learning algorithms for classification, recognition and diagnosis. Currently, we have projects on biometrics as well as diagnosis of epilepsy and traumatic brain injury (TBI) and the students can participate on those on-going direction.

Project #3 – Wireless Sensing Systems

Description:

We build systems with inductive wireless power transfer and sensing components. This group of projects include these directions:

1. Wireless charging to small appliances
2. Battery-less wireless implants
3. Battery-less wireless sensing, for example for pH levels inside a small tube

Project #4 – Zebrafish Electrophysiological Monitoring

Description:

Our group pioneers in the development of systems for electrophysiological assessment throughout the life of zebrafish. These projects include

1. Development of apparatus systems for ECG&EEG recording from multiple fish
2. Systems for multi-modal assessment of zebrafish embryos during developmental stages

Project #1: Python-based High-Level Programming Flow for CPU-FPGA Heterogeneous Systems

Description:

The fast-growing complexity of new applications and new use scenarios, e.g., deep learning, graph analytics, etc., poses serious challenges for computing systems. Heterogeneous systems consist of different types of processors and accelerators and provide unique combined benefits of hardware acceleration from each individual component. CPU-FPGA heterogeneous systems provide both programmable logic and general-purpose processors, and they have demonstrated great flexibility, performance, and efficiency. Heterogeneous systems have been created and deployed in many different applications and scenarios. However, as system complexity and application complexity grow rapidly, programming and optimizing heterogeneous systems require great manual efforts and consume a lot of time. In this project, we will develop a Python-based high-level programming framework on top of our PyLog flow to simplify programming and optimization of CPU-FPGA heterogeneous systems. The proposed high-level operations isolate underlying hardware details from programmers and provide more optimization opportunities for the compiler.

Student’s Involvement and Expected Outcomes:

This project has three stages. In the first stage, the student will get familiar the whole design flow and build CPU-FPGA accelerator prototypes using the Python-based high-level synthesis (HLS) tool, PyLog. In the second stage, the student will develop the compiler component within PyLog to enable accelerator system generation for CPU-FPGA heterogenous systems. In the third stage, the student will evaluate the compilation flow in terms of accelerator performance, energy efficiency, and programmability. 

Prerequisites: Python programming, Verilog/VHDL, FPGA design flow, knowledge in HLS or compiler is a plus.

Recommended Websites and Publications:

• Sitao Huang et al, PyLog: An Algorithm-Centric Python-Based FPGA Programming and Synthesis Flow, IEEE Transactions on Computers, Volume: 70, Issue: 12, Dec. 1, 2021. https://ieeexplore.ieee.org/document/9591456
• https://github.com/hst10/pylog
• Ryan Kastner et al, Parallel Programming for FPGAs. https://arxiv.org/abs/1805.03648

Project #2: Function as a Service (FaaS) for Serverless Heterogeneous Hardware Acceleration

Description:

Distributed Function as a Service (FaaS) platforms enables scalable and flexible remote function execution. Recently, FaaS based serverless architecture is getting more and more attention in cloud computing and high-performance computing domains, and has been used to accelerate applications in machine learning, scientific computing, cloud services, etc. As hardware accelerators like GPUs and FPGAs becoming pervasive in both cloud ends and edge ends, leveraging the computing capabilities of these accelerators is becoming a major challenge. This project aims to enable Function as a Service (FaaS) for hardware accelerator platform by building a Python-based high-level FaaS programming flow.

Student’s Involvement and Expected Outcomes:

This project has three stages. In the first stage, the student will get familiar the whole design flow and build FaaS prototype platforms using the FuncX. In the second stage, the student will develop the compilation flows with FuncX to enable automated FaaS programming for heterogenous systems. In the third stage, the student will evaluate the whole development flow in terms of computing performance, energy efficiency, and programmability. 

Prerequisites: Python programming, knowledge in hardware design (Verilog/VHDL) or compiler is a plus.

Recommended Websites and Publications:

• https://funcx.org/
• Sitao Huang et al, PyLog: An Algorithm-Centric Python-Based FPGA Programming and Synthesis Flow, IEEE Transactions on Computers, Volume: 70, Issue: 12, Dec. 1, 2021. https://ieeexplore.ieee.org/document/9591456
• https://github.com/hst10/pylog
• Ryan Kastner et al, Parallel Programming for FPGAs. https://arxiv.org/abs/1805.03648

Project #3: Machine Learning based Hardware Accelerator Modeling and Optimization

Description:

The exploding complexity and computation efficiency requirements of applications are stimulating a strong demand for hardware acceleration with heterogeneous platforms such as FPGAs and GPUs. Recently, many new hardware accelerators have been introduced, e.g., NVIDIA Tensor Cores, Xilinx AI engines in Versal ACAPs, Google TPUs, etc. The hardware complexity of these accelerators is growing rapidly. However, a high-quality accelerator design is very hard to create and optimize as it requires hardware acceleration expertise and a long design iteration time. This project aims to mitigate the modeling and optimization challenge of hardware accelerators using machine learning techniques.

Student’s Involvement and Expected Outcomes:

This project has three stages. In the first stage, the student will get familiar the whole hardware acceleration design flow and build prototype accelerator platforms. In the second stage, the student will develop modeling and optimization flows using existing compilation infrastructures to enable automatic performance evaluation and modeling. In the third stage, the student will build the optimization flow for accelerator design and evaluate the whole development flow in terms of computing performance, energy efficiency, and programmability.

Prerequisites: machine learning,C/C++/Python programming, knowledge in hardware design (Verilog/VHDL), GPU programming, or compiler is a plus.

Recommended Websites and Publications:

• Mang Yu et al, Chimera: A Hybrid Machine Learning-Driven Multi-Objective Design Space Exploration Tool for FPGA High-Level Synthesis, IDEAL 2021. https://link.springer.com/chapter/10.1007/978-3-030-91608-4_52
• Lianmin Zheng, et al, Ansor : Generating High-Performance Tensor Programs for Deep Learning. https://arxiv.org/abs/2006.06762

Kartik Hegde et al, Mind mappings: enabling efficient algorithm-accelerator mapping space search. ASPLOS 2021. https://dl.acm.org/doi/10.1145/3445814.3446762

Dr. Shen’s research interests include machine learning, Bayesian methods, sports analytics, biostatistics, and causal inference.

Project #1: Large Language Models in Sports Analytics

Description:

Multimodal Large Language Models (MLLMs) are pushing the boundaries of AI in sports analytics by enabling more advanced reasoning about complex sports scenarios through the integration of both textual and visual information. This project aims to build a large-scale benchmark dataset designed to evaluate the performance of MLLMs across a variety of multi-level sports reasoning tasks. These tasks will include foul detection, rule application, play strategy analysis, and event prediction. The hope is to improve machine’s understanding and decision-making in sports scenarios, potentially leading to better tools for coaching, refereeing, and sports commentary.

Students’ Involvement and Expected Outcomes:

Students will participate in all aspects of the project, including literature review, data collection, model tuning and evaluation, and manuscript writing. Under the faculty mentor and graduate student’s supervision, students will gain hands-on experience on language models, video processing and statistical analysis. 

Prerequisites:

Passionate about sports analytics. Good understanding of large language models and data analysis. 

Project #2: Statistical Modeling of Path Data in Minecraft

Description:

This project aims to develop new statistical methodologies to enhance our understanding of the mechanisms underlying human spatial memory and navigation in 3D environments. By analyzing movement path data collected from a Minecraft game environment, we plan to address a series of questions such as how do humans form and retain precise spatial relationships in a 3D environment, and how do environmental features and landmarks influence navigation and spatial memory.  

Students’ Involvement and Expected Outcomes:

Students will participate in all aspects of the project, including literature review, data cleaning and analysis, model implementation and tuning, and manuscript writing. Under the faculty mentor and graduate student’s supervision, students will gain hands-on experience on statistical analysis, machine learning and data visualization. 

Prerequisites:

Good understanding of statistics and machine learning. Coding experience in R or Python. 

Project #1: Towards Trustworthy Explanation: On Causal Rationalization

Description:

With recent advances in natural language processing, rationalization becomes an essential self-explaining diagram to disentangle the black box by selecting a subset of input texts to account for the major variation in prediction. Yet, existing association-based approaches on rationalization cannot identify true rationales when two or more snippets are highly inter-correlated and thus provide a similar contribution to prediction accuracy, so-called spuriousness. To address this limitation, we novelly leverage two causal desiderata, non-spuriousness and efficiency, into rationalization from the causal inference perspective. We formally define a series of probabilities of causation based on a newly proposed structural causal model of rationalization, with its theoretical identification established as the main component of learning necessary and sufficient rationales. The superior performance of the proposed causal rationalization is demonstrated on real-world review and medical datasets with extensive experiments compared to state-of-the-art methods.

Students’ Involvement and Expected Outcomes:

The student will be cross-trained to work together throughout the summer to conduct simulation studies on Large Language Models, and code development for the project and its extension. For simulation studies, the student will implement programs to simulate synthetic data, conduct simulation studies using the computation engine, and generate tables and figures for the project. For possible code development, the student will test the computation engine for accuracy, develop user-friendly Python/R/C packages based on the computation engine, test the package for robustness, prepare examples, and write user guides.

Prerequisites:

Python/R/C coding experience; at least one research experience before; and strong self-motivation.

Recommendation Websites and Publications:

– Paper: https://proceedings.mlr.press/v202/zhang23ap/zhang23ap.pdf

– Tutorial Book: https://causaldm.github.io/Causal-Decision-Making/Overview.html

–  Website: https://hengruicai.github.io/