Projects

Content Filters

Results

Active Visual Planning: Handling Uncertainty in Perception, Prediction, and Planning Pipelines

Adaptive Long-Distance Navigation for Autonomous Drones

Adding Safety and Robustness to Learning for Robots by Learning on Robots

Addressing Challenges in Large-scale Distributed AI Systems

Alpa: A Distributed System for Training and Serving Large Models

Amazon-Berkeley Objects: A Large-Scale Dataset for 3D Object Understanding

Animating Animals from Video

Ashera: Neural Optimization Modulo Theories

Automated Collision Prediction in Autonomous Systems with Monocular Camera

Automated State and Action Space Design for Multi-Objective Reinforcement Learning

Automating Multi-Agent Curriculum Learning with Probabilistic Programs

Coherent and Consistent Long Story Generation

Combating Hallucination in Conditional Sequence Generation

Compressing High Capacity Models with Implicit Neural Networks and Frank-Wolfe

Control of Microrobots with Data and Computationally Efficient Reinforcement Learning

Data Curation for Web-Scale Datasets

Designing Pro-social Networks through Multi-Agent Learning Theory

Dimension-free Statistical and Computational Guarantee for Optimal Transport

Distributed Learning: Privacy and Data Summarization

Dynamic Compression Techniques for Efficient Transformers

Emergent Collaboration for Heterogeneous Multi-Robot Rearrangement

Enabling Non-Experts to Annotate Complex Logical Forms at Scale

Experimenting with Zero-Knowledge Proofs of Training

Factorized language representations with knowledge and logic

Fate of Snow

Formal Skill Representation and Composition for Task Generalization

Generalizing Domain Randomization for Zero-Shot Transfer

Grounded and Modular Vision and Language Learning

Grounded and Structured Self-Supervised Pre-training of Speech for Spoken Language Model

GuBERT: Grounded units for Self-Supervised Pre-training of Speech

Hardware Software Co-Design for NLP and Recommendation Systems

Hierarchical Model-Based Reinforcement Learning with Temporal Abstractions

Interactive Learning from Vision and Touch

Investigations into the Complexity of Nonconvex Optimization

Large-scale 3D Reconstruction from Multi-view Image Datasets

Learning Dexterous In-Hand Manipulation with Vision and Touch

Learning From Play in Children and Robots: Who Can Train a Robot Better?

Learning human-robot collaboration from human feedback

Learning Large Touch-Vision-Language Models Using Self-Supervised Robot Learning

Learning Safety-Assured Collaborative Quadrupedal Manipulation

Learning Selective Invariance Upon Parametric Density Functions of Transformations

Learning Space Partitions for Path Planning

Learning Successor Affordances as Temporal Abstractions

Learning to Collaborate with Human Players

Learning-based Safe Navigation for Dynamic Legged Robots

Learning-Driven Exploration For Search

Leveraging Demonstrations with Goal-Directed Multi-Task Bisimulation

Local Parametric Learning Rules for Parallel and Scalable Training

Long Term Video Understanding

Long-Range Understanding and Consistency in Story Generation

Low-Data Learning for Assistive Video Description

ML-Based Robotic Manipulation via the Use of Diverse Datasets

Modeling Interpersonal Multimodal Signals in Social Conversation

Modeling latent variable for self-supervised learning

Multi-agent Social Learning

Multiscale Modeling for Control

Never Decrypt Data Lake

NumS: NumPy API-Compatible Framework backed by Ray

Offline Recovery RL: Offline Reinforcement Learning with Safe Online Adaptation

Optimal Data Augmentation Strategy Search

Personalized federated learning: new algorithms and statistical rates,

Pre-trained Representations for Language-Guided Web Navigation

Realistic Large-Scale Benchmark for Adversarial Patch

Regret Bounds for Contextual Bandits Under Slate Feedback

Reinforcement Learning in High Dimensional Systems

Robust Image Classification via Parts and Disentangled Attributes

Robustness for Deep Learning/Ethical AI Through Human Value Modeling

Safe and Sound: Learning Locomotion Skills Across Robot Morphology

Safe Robotic Learning Via Reachability Theory

Secure and Privacy-Preserving Federated Learning

Self Supervised Semantic Segmentation in the Wild

Self-supervised Learning for Generic Visual Representation

Self-supervised Open-World Segmentation

Statistically Efficient Offline RL with General Function Approximation

Task-Specific World Models for Robotic Manipulation

Towards a Unified Understanding of Privacy and Generalization for Better Algorithm Design

Towards Human-like Attention

Towards Learning and Auditing Private Foundation Models

Towards Robust Neural Networks with Conditional Generative Models

Training Sparse High Capacity Models with Implicit Neural Networks and Frank-Wolfe

Uncertainty Aware Machine Learning for Model Based Planning and Control

Universal Representation Learning for Control

Unsupervised Environment Design for Multi-task Reinforcement Learning

Using Deep Reinforcement Learning to Generalize Search in Games

Video Representation Learning for Global and Local Features

Weakly Supervised Multimodal Feature Representation Learning from Video

Writing with Speech — Using LLMs for Gist-Level Manipulation of Spoken Text

X-Ray for Lateral Access Mechanical Search