Do Pretrained Transformers Learn In-Context by Gradient Descent?
Explore the relationship between In-Context Learning and Gradient Descent in large language models, examining theoretical connections and empirical evidence from pre-trained transformers.
Localization vs. Semantics: Visual Representations in Unimodal and Multimodal Models
Explore visual representations in unimodal and multimodal models, comparing their strengths in label prediction and dense prediction tasks. Gain insights into language's role in visual learning.
Prompting Language Models Improves Quoting from Pre-Training Data - EACL 2024
Explore how "according-to" prompting enhances LLMs' ability to ground responses in factual data, reducing hallucinations and improving task performance across diverse corpora.
Clean Label Poisoning Attacks: From Classification to Speech Recognition
Explore poisoning attacks in speech recognition, from adversarial manipulations to innovative defenses. Uncover strategies, mechanisms, and future research in AI security.
Joint Energy-based Models for Speech
Explore generative models, focusing on Energy-based Models and Joint Energy-based Models for speech, with potential applications in multiple tasks.
Model Distillation for Faithful Explanations of Medical Code Predictions
Explore knowledge distillation techniques for generating faithful explanations of medical code predictions, enhancing trust in machine learning models for clinical decision-making.
A Hyperparameter Optimization Toolkit for Neural Machine Translation Research
Explore a toolkit for optimizing hyperparameters in neural machine translation research, enhancing efficiency and performance in NMT model development.
Causal Analysis and Decision Intelligence for Manufacturing at Bosch
Explore neuro-symbolic AI applications in manufacturing, focusing on causal analysis and decision intelligence for improved processes at Bosch. Learn about root-cause analysis and cognitive architectures for decision-making.
Intersection of Physics and AI - Incorporating Physics into Data-Driven Computer Vision
Explore the fusion of physics and AI, covering applications in NLP, healthcare, and computer vision. Learn innovative approaches to incorporate physics into data-driven models.
Generative AI for Creative Design of Novel Materials
Explore generative AI techniques for innovative material design, focusing on creative approaches to develop novel materials with desired properties and functionalities.
Civilizing AI: Examining Emerging Capabilities and Mitigation
Explore emerging AI capabilities and mitigation strategies with Dr. Amitava Das. Gain insights into civilizing AI and its implications for future technological advancements.
Enhancing Healthcare with AI-in-the-Loop
Explore AI's role in healthcare through rich data learning, human-AI collaboration, and expert advice integration. Discover applications in electronic health records, clinical studies, and surveys.
Adversarial AI/ML Issues and Defense Strategies
Explore adversarial AI/ML challenges and defense mechanisms with Dr. Dipankar Dasgupta, gaining insights into critical cybersecurity strategies for machine learning systems.
PhD Students Overview - Part 3
Explore advanced research topics and methodologies for doctoral studies in AI, focusing on Dr. Sheth's expertise and current trends in the field.
PhD Students Overview - Part 2 - Lecture 3
Explore the research contributions of Dr. Sheth's past Ph.D. students in this comprehensive overview of their work and achievements.
PhD Students Work Overview - Part 1 - Lecture 2
Explore groundbreaking research by Dr. Sheth's former Ph.D. students, gaining insights into innovative AI projects and their impact on the field.
Introduction to Knowledge Graphs and Semantic Technologies - Lecture 1
Explore foundational concepts in knowledge representation, databases, and semantic technologies. Learn about schema integration and semantic proximity in data management.
AI for Credible Elections
Explore AI's role in ensuring credible elections through expert discussions on challenges, solutions, and ethical considerations in electoral processes.
Rebooting AI: Building Artificial Intelligence We Can Trust - Book Review Part 2
Explore key insights from "Rebooting AI" in this book review, examining current AI limitations and future potential for more robust and reliable artificial intelligence systems.
Neuro-Symbolic AI for Mental Healthcare
Explore neuro-symbolic AI methods for enhancing mental healthcare interventions, focusing on social media data analysis, clinical knowledge integration, and improved conversational systems.
Knowledge-infused Learning for Autonomous Driving (KL4AD)
Explore knowledge-infused learning techniques for autonomous driving, focusing on neuro-symbolic AI and its potential to enhance interpretability, explainability, and robustness in real-world applications.