Deployable AI (DAI)

                                                Tentative Schedule

Monday 26 February 2024

9:00 am to 9:05 am           Opening Remarks (5 mins)

9:05 am  to 9:40 am             Invited Speaker 1: Sunipa Dev, Research Scientist, Google Research

9:40 am to 10:25 am             Oral Presentations Session-1

10:25 am  to 10:35 am     Spotlight Presentation Session-1 

10:35 am to 11:00 am     Break and Poster Session-1

11:00 am to 11:30 am             Oral Presentations Session-2  

11:30 am to 12:05 pm             Invited Speaker 2: Srijan Kumar, Assistant Professor, College of Computing at Georgia Institute of Technology

12:05 pm to 1:45 pm             Lunch Break

1:45 pm to 2:20 pm             Invited Speaker 3: Pranesh Srinivasan, Senior Engineer Google

2:20 pm to 2:55 pm             Invited Speaker 4: Xiaoxiao LI, Assistant Professor, University of British Columbia 

2:55 pm to 3:25 pm                 Oral Presentations Session-3  

3:25 pm to 3:35 pm                 Spotlight Presentations Session-2

3:35 pm to 4:00 pm             Break and Poster Session-2

4:00 pm to 4:45 pm             Oral Presentations Session-4  

4:45 pm to 4:55 pm                 Spotlight Presentations Session-3

4:55 pm to 5:00 pm                 Closing Remarks

                                                                      Tentative Detailed Schedule

For Invited Speakers, We will have 30 mins of Presentation and 5 mins of Q&A

Oral Presentations Session-1 (12 mins+3 mins (Q&A)) 9:40 am to 10:25 am

9:40 am to 9:55 am Unraveling the Temporal Dynamics of the Unet in Diffusion Models

09:55 am to 10:10 am MABViT - Modified Attention Block Enhances Vision Transformers

10:10 am to 10:25 am Preliminary Investigations of a Multi-Faceted Robust and Synergistic Approach in Semiconductor Electron Micrograph Analysis: Integrating Vision Transformers with Large Language and Multimodal Models

Spotlight Presentations Session-1 (2 mins each) 10:25 am to 10:35 am

Unveiling Empirical Pathologies of Laplace Approximation for Uncertainty Estimation

Towards Reliable AI Model Deployments: Multiple Input Mixup for Out-of-Distribution Detection

Credibility-aware Reliable Multi-Modal Fusion Using Probabilistic Circuits

EEE-QA: Exploring Effective and Efficient Question-Answer Representations

        Reprogramming Foundational LLMs for Enterprise Adoption for Spatio-Temporal    Forecasting Applications: Unveiling a New Era in Copilot-Guided Cross-Modal Time Series Representation Learning

Oral Presentations Session-2: 11:00 am to 11:30 am

11:00 am to 11:15 am Closing the Gap in the Trade-off between Fair Representations and Accuracy

11:15 am to 11:30 am         Elevating Defenses: Bridging Adversarial Training and Watermarking for Model Resilience

Poster Session-1 10:35 am to 11:00 am

All the papers scheduled to be presented in the morning session (Oral/Spotlight) will have a poster session during this session.

Oral Presentations Session-3 (12 mins+3 mins (Q&A)) 2:45 pm to 3:15 pm

2:40 pm to 3:00 pm MVSA-Net: Multi-View State-Action Recognition for Robust and Deployable Trajectory Generation

3:00 pm to 3:15 pm Deployable Reinforcement Learning with Variable Control Rate

Spotlight Presentations Session-2 (2 mins each) 3:15 pm to 3:35 pm

vPALs: Towards Verified Performance-aware Learning System For Resource Management

Improved Decision Module Selection for Hierarchical Inference in Resource-Constrained Edge Devices

Composing Efficient, Robust Tests for Policy Selection

Uncertainty-aware Language Modeling for Selective Question Answering

         Large Language Model Inference with Lexical Shortlisting

Oral Presentations Session-4 (12 mins + 3mins (Q&A)) 4:00 pm to 4:45 pm

        2:00 pm to 2:15 pm Directed Structural Adaptation to Overcome Statistical Conflicts and Enable Continual Learning

        2:15 pm to 2:30 pm  Half-Space Feature Learning in Neural Networks

        2:30 pm to 2:45 pm  Advancing Ante Hoc Explainable Models through Generative Adversarial Networks

Spotlight Presentations Session-3 (2 mins each) 4:45 pm to 4:55 pm

Learning Credal Conditional Probability Tables with Qualitative Knowledge

Maintaining User Trust Through Multistage Uncertainty Aware Inference

        Plug and Play with Prompts: A Prompt Tuning Approach for Controlling Text Generation

        REVAMP: Automated Simulations of Adversarial Attacks on Arbitrary Objects in Realistic Scenes

Using Concept Embeddings to Understand Neural Net Training

Poster Session-2   3:35 pm to 4:00 pm

All the papers scheduled to be presented in the afternoon session (Oral/Spotlight) will have a poster session during this session.