🏢 Massachusetts Institute of Technology

Latent Graph Diffusion (LGD) unifies graph learning, solving all task levels and types with a single framework and state-of-the-art results.

Heterogeneous Pre-trained Transformers (HPT) enables robots to learn generalizable policies from diverse data, drastically improving performance on unseen tasks.

Revolutionizing image generation, Representation-Conditioned Generation (RCG) achieves state-of-the-art results in unconditional image synthesis by leveraging self-supervised representations to condit…

Single image to physically compatible 3D objects: A new framework ensures 3D models maintain stability and mirror real-world equilibrium states, advancing realism in dynamic simulations and 3D printi…

SGD’s dynamics are precisely characterized by the interplay of noise and symmetry in loss functions, leading to unique, initialization-independent fixed points.

Auto-Regressive Decision Trees (ARDTs) surprisingly outperform Transformers on language tasks!

MDAgents: An adaptive multi-agent LLM framework boosts medical decision-making accuracy by dynamically adjusting collaboration structures based on task complexity.

DMD2 dramatically speeds up image generation by cleverly distilling expensive diffusion models, achieving state-of-the-art results without sacrificing quality.

Boost AI predictions by using human judgment on algorithmically indistinguishable inputs!

Linear gated neural networks with fast, bounded units self-organize into modular weight structures and unique gating representations, enabling flexible task switching and compositional generalization.

Boosting preference learning, this research uses human response times to improve linear bandit algorithms, significantly accelerating preference identification.

Double-Ended Synthesis Planning (DESP) significantly boosts computer-aided synthesis planning by using a bidirectional search, outperforming existing methods on multiple benchmarks, especially when sp…

Autoregressive image generation is revolutionized by eliminating vector quantization, achieving strong results with increased speed using a novel diffusion procedure.

This paper presents a novel unified framework for deriving information-theoretic lower bounds for bandit learnability, unifying classical methods with interactive learning techniques and introducing a…

Graph Neural Networks (GNNs) learn optimal approximation algorithms for combinatorial optimization problems, achieving high-quality solutions for Max-Cut, Min-Vertex-Cover, and Max-3-SAT, while also p…

A novel machine learning recipe drastically accelerates charge density prediction in density functional theory, achieving state-of-the-art accuracy while being significantly faster than existing metho…