Eight YWCC Research Projects Funded Through AI@NJIT Initiative
Seven faculty members in the Ying Wu College of Computing (YWCC) have received grant support to fund eight research projects as part of the Grace Hopper Artificial Intelligence Institute at NJIT (GHRI). The Institute was launched through a $6 million investment by an anonymous donor and matching funds in collaboration with the university’s $10 million AI@NJIT initiative.
Funded proposals focus on any aspect of AI, including but not limited to machine learning, natural language processing, robotics, AI ethics, and applications of AI across various disciplines.
“The Grace Hopper AI Research Institute exemplifies our dedication to innovation and interdisciplinary collaboration, positioning our university as a leader in AI research and application,” said John Pelesko, provost and senior executive vice president, in a February article announcing the initial launch.
YWCC Dean Jamie Payton added, "The seven faculty members chosen to represent GHRI, the college and the university are leaders in their respective fields, pioneering AI research and discovery with wide-ranging impact—including in safety and security, health care, and even space weather, which has effects on Earth's atmosphere and our communication infrastructure."
Associate Professor Hai Phan (Data Science) has received AI@NJIT funding for two research proposals:
FunSec: Certified Functional and Secure Code Generation with Large Language Models:
The project aims to develop a novel AI assistant for software developers that generates Functional & Secure code snippets together with a rigorous probabilistic guarantee with a confidence interval a for the functionality and security of the generated small snippets. The research will mitigate significant security vulnerabilities in CLLM-generated (Code LLMs) code, which pose severe risks, such as data breaches, unauthorized access to sensitive information, identity theft, financial loss, malware infection, disruption of operations, etc., when deployed in real-world software systems.
Automating Chip Design with AI: The Journey from Zero to Silicon:
This project aims to develop the first comprehensive LLM-based ecosystem, enabling an effective pipeline for automating AI accelerator generation and verification, ensuring a seamless transition from high-level specifications to low-level implementations considering budgetary restraints (e.g., power, latency, area). This research is well-aligned with the CHIPS Act to strengthen domestic semiconductor manufacturing, design and research.
Distinguished Professor David Bader (Data Science)
AI-Enhanced Graph Pattern Matching and Cluster Analysis for Neural Circuit Discovery:
This project advances fundamental neuroscience research to understand the complexity of brain networks and their functional organization. The research will develop novel AI-enhanced algorithms and systems that can recognize both functionally similar neural patterns and significant network clusters through multimetric analysis, directly supporting neuroscientists in their quest to understand the brain’s connectome. The innovation utilizes the research team’s established expertise in parallel graph algorithms and large-scale dating processing.
Professor Chengjun Liu (Computer Science)
Toward Increased Rosacea Awareness Among Population Using Advanced AI: Explainable AI Automatic Rosacea Diagnosis and Region of Interest Detection:
The project utilizes deep learning and explainable statistical approaches to increase rosacea awareness, which afflicts approximately 16 million Americans according to the National Rosacea Society. The results will better assist physician diagnosis on the disease using interpretable automatic detection methods, with three-fold contributions. The first can automatically distinguish patients suffering from rosacea from people who are clean of this disease with a significantly higher accuracy than other methods in unseen test data, including both classical deep learning and statistical methods. The second addresses the interpretability issue by measuring the similarity between the test sample and the means of two classes, namely the rosacea class versus the normal class, which allows medical professionals and patients to understand and trust the results. The final contribution will not only help increase awareness of rosacea in the general population but will also remind patients who suffer from the disease of possible early treatments, as rosacea is more treatable in its early stages.
Assistant Professor Akshay Rangamani (Data Science)
Does it Add Up? Characterizing Mathematical Skills in Language Models
This project aims to characterize the parameter geometry in language models trained on mathematical problems. This characterization will help the research team understand a) when language models go beyond pattern matching to learn generalizable representations, b) how language models learn hierarchical skills, and c) how model capabilities grow with model size. While the project will focus on the controlled domain of arithmetic and algebra, the team will aim to generalize their findings to more general language models. Beyond the intellectual merit, this project will also train undergraduate students to perform foundational AI research, use the latest AI models, and allow them to interact with the broader research community through conferences and workshops.
Professor Jason Wang (Computer Science)
Prediction of Extreme Events in Space Weather Using Generative Artificial Intelligence and Multimodal Machine Learning Techniques:
Research will entail development of an AI toolbox for multimodal space weather (SWx) forecasting. The toolbox will integrate data spanning three solar cycles, enabling a comprehensive analysis, interpretation and prediction of solar transient events and their precursors. A key innovation of the project is the creation of novel datasets using generative AI techniques. The researchers envision that these enhanced datasets will provide new insights into the dynamics of solar active regions and their impact on space weather. The results will significantly enhance the ability to forecast extreme SWx events, including solar flares, coronal mass ejections and solar energetic particle events, and mitigate their effects on technological systems.
Assistant Professor Lijing Wang (Data Science)
Large Language Model-Driven AI Platform for Next-Generation Surgical Planning and Navigation
The project will develop a next-generation AI-based platform for automated surgical planning and navigation through incorporation of a large language model with state-of-the-art machine learning algorithms in image processing. It will further prepare surgery-critical information in 3D augmented reality (AR) space for intuitive visualization. The platform has the potential to help neurosurgeons protect functional eloquent regions of the brain, control surgical risk and improve patient outcomes. Expected results will fully automate the surgical preparation process and release the power of surgical teams. The success of this project will improve the value proposition for next-level surgical innovation and result in patentable technologies for NJIT.
Assistant Professor Mengjia Xu (Data Science)
Predicting Solar Active Regions with the Aid of Artificial Intelligence
The proposed project will leverage advanced sequence models to bridge significant knowledge gaps in solar active regions (AR) prediction and evolution, offering a transformative approach to forecasting space weather while addressing the need for extended prediction windows to meet the operational requirements of modern space missions. The research has two objectives: 1) Train two different deep learning models on extended solar AR emergence datasets. 2) Develop an open-source machine learning dataset visualization tool for AR emergent precursors. The expected outcomes will aid with the safety and success of NASA’s space-based operations by mitigating solar flares, coronal mass ejections and solar energetic particles. Such events can potentially downgrade radio communications, incapacitate satellites, expose airline passengers to elevated radiation levels and endanger life in outer space.