STARTER-AI Workshop Summer 2025, supported by NSF

May 21 - 23, 2025

Day 1: Opening & AI Research Highlights

The workshop opens with welcoming remarks and invited talks from external and UTRGV speakers. The afternoon features a showcase of AI research projects, including oral and poster presentations from students and faculty across UTRGV campuses.

Day 2: Hands-on AI Tutorials

Participants will take part in a series of hands-on sessions designed to introduce foundational tools and techniques in AI. Topics will include working with AI frameworks, building simple models, and exploring applications of language models. Link to Repository of Workshop Materials

Day 3: Research Infrastructure & Collaboration

The final day will feature an overview of UTRGV’s high-performance computing resources for AI research and access to national resources to advance your AI research. Link to Repository of HPC MaterialsIntro to HPC Slides (PPT)

 

Locations

Schedule

Day Time Event
Day 1:

May 21
9:00 AM Check-in / Registration / Coffee
9:45 AM Welcome, Chair: Dr. Soumya Mohanty
10:00 AM UTRGV Speaker - Dr. Sanjeev Kumar
10:50 AM Break
11:00 AM AI@UTRGV Showcase Oral Presentations, Chair: Dr. Sanjeev Kumar
Part 1
Noon Lunch
1:30 PM AI@UTRGV Showcase Oral Presentations, Chair: Dr. Sanjeev Kumar
Part 2
2:30 PM Distinguished Speaker - Dr. Robert Luo from UNM
3:30 PM Break
3:45 - 4:30 PM AI@UTRGV Research Poster Showcase
Posters: Edinburg, Brownsville
Day 2:

May 22
Theme: Beginner-Friendly, Hands-on AI Workshops with PyTorch in Google Colab
9:00 AM Check-in / Coffee
9:30 AM UTRGV Speaker - Dr. Francis Kofi Andoh-Baidoo
10:30 AM Intro to Google Colab + Setup
Walkthrough of Colab, notebook runtime, and GPU setup
Launch Notebook
10:50 AM Break
11:00 AM Mini Workshop 1: Your First Neural Network
Intro to tensors, gradients, and training a simple classifier
Launch Notebook
Noon Lunch
1:30 PM Mini Workshop 2: Sentiment Classification
Train a PyTorch text classifier on short reviews (positive/negative)
Launch Notebook
2:30 PM Break
2:45 - 3:45 PM Mini Workshop 3: Fine-Tuning a Pretrained Language Model
Use Hugging Face's DistilBERT to fine-tune on a small sentiment dataset
Launch Notebook
Bonus Capstone: Fine-Tune on Emotion Dataset
Day 3:

May 23
9:00 AM Check-in / Coffee
9:30 AM Session: Getting Started on Cradle HPC (Part 1)
Chair: Dr. Erik Enriquez
10:50 AM Break
11:00 AM Session: Getting Started on Cradle HPC (Part 2) and Closing Remarks
12:00 PM Workshop Concludes

 

Section Chairs and Invited Speakers

 

Invited Speakers

Dr. Xin Luo

Merging AI with Business Practice: Two Projects for Investment and Healthcare Industries

Xin (Robert) Luo, Ph.D.

Professor of Management Information Systems at Robert O. Anderson School of Management

University of New Mexico, Albuquerque, NM

Bio:

Dr. Xin (Robert) Luo is a Distinguished Professor of Management Information Systems and an Endowed Dean's Professor of Research Excellence at the Robert O. Anderson School of Management of The University of New Mexico in Albuquerque, New Mexico, USA. He is ranked No.1 worldwide for publication in the AIS Senior Scholars List of Premier Journals for 2022–2024. Additionally, he has been listed in the Top 2 % Global Scientists list published by Stanford University and Elsevier. His research interests center around behavioral information systems security management, consumer privacy protection, e-commerce, innovative technologies for business strategic decision-making and development, and cross-cultural IT management. He has published in leading journals, including Information Systems Research, Journal of Operations Management, Production and Operations Management, Journal of Management Information Systems, Journal of the Association for Information Systems, European Journal of Information Systems, Information Systems Journal, Journal of Strategic Information Systems, Journal of Information Technology, Decision Sciences, Decision Support Systems, and Information & Management. He currently serves as a senior editor for Decision Support Systems and Internet Research and as an associate editor for the Journal of the Association for Information Systems, Information & Management, Electronic Commerce Research, and Journal of Electronic Commerce Research. He sits on the Editorial Review Board of Information Systems Research. He is the co-editor-in-chief of the International Journal of Accounting and Information Management.


Dr. Sanjeev Kumar

AI & Security for the Connected World

Dr. Sanjeev Kumar

Professor, Department of Electrical & Computer Engineering

Founder of NSF funded Cyber Security Research lab

University of Texas Rio Grande Valley, Edinburg, TX 78539

Abstract:

AI is increasingly being used to combat evolving cyber threats, vulnerabilities and attacks to secure today's connected world involving IoT devices, networks, and data. AI is also being used by attackers to launch more sophisticated attacks, hence AI-driven defense is even more important today. In this talk, we will discuss how AI is being used on both sides, by attacks and defenders, and how it can impact the future of our connected world.

Bio:

Dr. Kumar is a Professor in the Department of Electrical and Computer Engineering, UT-RGV. He has been a Faculty Fellow, Houston Endowment Chair for Science, Math and Technology, 2017-2018; Faculty Fellow, Lloyd M. Bentsen, Jr. Endowed Chair in Engineering 2018-2020. He has over 30 years of industry and academic research and training experience. Dr. Kumar received the PhD degree in Computer Engineering/ ECE dept. from North Carolina State University, Raleigh, NC. He has published over 80 peer-reviewed articles in Journals/Book Chapters/Conference proceedings. He is a founder/director of NSF funded -Cyber Security Research Lab at UT-RGV. He is a winner of prestigious Univ. Of Texas Regents’ Outstanding Teaching Award and many UTRGV/UTPA teaching excellence awards. He has received funding from NSF, NIH and the Department of Homeland Security. He is a senior member of IEEE.


Dr. Francis Kofi Andoh-Baidoo

Evolution of AI

Dr. Francis Kofi Andoh-Baidoo

Professor, Department of Information Systems

University of Texas Rio Grande Valley, Edinburg, TX 78539

Bio:

Dr. Francis Kofi Andoh-Baidoo is Professor in the Department of Information Systems. He received his Doctor of Philosophy degree in Business Administration with a major in Information Systems and a minor in Finance from Virginia Commonwealth University. He also has Masters in Business Administration and Masters in Information Technology Management from the University of North Carolina at Greensboro. His undergraduate in Materials Engineering is from the Kwame Nkrumah University of Science and Technology, Ghana. He serves on the editorial review board of several Information Systems journals. He is a certified Oracle Database Administrator and has industry experience in Enterprise Resource Planning Systems and Business Intelligence.


 

AI@UTRGV Research Showcase Presentations

Approximating Deep Speech Models with Transparent Time Series Techniques

Author(s): Gaukhar Nurbek, Richard Tapia, Brooklyn Berry, Juan Manuel Perez

Abstract: Mispronunciations are common among non-native English speakers and pose challenges for speech-based models. In this work, we aim to understand how an existing deep learning time series classifier makes its decisions. We train LITETime model on a controlled dataset of correctly and incorrectly pronounced words, and evaluate it using structured perturbations generated by our custom SquareSpecAugment augmentation method. To interpret LITETime’s decision behavior, we design a transparent model using Dynamic Time Warping (DTW) and k-Nearest Neighbors (k-NN) that replicates LITETime’s predictions. Our experiments show that this simple DTW + k-NN model can match LITETime’s outputs with 100% agreement, both on augmented and original test samples. These findings suggest that LITETime’s decisions can be explained by surface-level similarity over Wav2Vec embeddings, and that interpretable models can effectively capture its behavior.


Adapting Multi-Agent Transformers for Competitive Scenarios in Multi-Agent Reinforcement Learning

Author(s): Daniel Masamba, Chen-Yuan Wang, Erik Enriquez, Dongchul Kim

Abstract: We extend Multi-Agent Transformers (MAT) to adversarial settings, focusing on the Simple Tag scenario in the Multi-Particle Environment (MPE). By partitioning policies and observations between opposing agents, our approach enhances adversarial training and improves learning. Results highlight the potential of transformer-based architectures in competitive MARL.