3D flash synapses for in-memory path planning in autonomous systems Self-driving cars need two things at once: vision to read the world (convolutions on images) and decision-making to choose where to go next (reinforcement learning). Today, those steps are usually run on

Genolator: A Multimodal Large Language Model Fusing Natural Language, Genomic, and Structural Tokens for Protein Function Interpretation 1. Genolator is a novel multimodal large language model that integrates natural language with genomic and protein structural data to interpret

RxnCaption: Reformulating Reaction Diagram Parsing as Visual Prompt Guided Captioning 1. The RxnCaption framework revolutionizes the parsing of chemical reaction diagrams by transforming it into an image captioning task, leveraging the power of Large Vision-Language Models

Physics-Diffusion-Driven Multiscale Aggregation for Drug-Target Interaction Prediction 1. A novel framework PDDMA-DTI is introduced to predict drug-target interactions with enhanced accuracy and robustness. This method leverages physics-inspired diffusion processes to capture

SMF-VO: Direct Ego-Motion Estimation via Sparse Motion Fields Sangheon Yang, Yeongin Yoon, Hong Mo Jung, Jongwoo Lim tl;dr: sparse optical flow->linear and angular velocity; generalized 3D ray-based motion field->different camera models https://t.co/NDQ1qDKvuk

Singer and Rotstein et al., "Time-to-Move: Training-Free Motion Controlled Video Generation via Dual-Clock Denoising" Make a rough warp, push it through Image-to-Video model with denoise together up until a timestep, then let it finish the rest without interference.

STAR-VAE: Latent Variable Transformers for Scalable and Controllable Molecular Generation 1. The vast chemical space of drug-like molecules necessitates powerful generative models. STAR-VAE addresses this by combining a Transformer encoder and autoregressive Transformer decoder,

Inside a real-time 3D mapping system! 🧭 That's how modern home bots map and localize using only cameras. @maticrobots is using voxel-based neural networks running on NVIDIA Jetson Orin to build real-time, photorealistic 3D maps of the world around its robots. Its autonomy

qcGEM: a graph-based molecular representation with quantum chemistry awareness 1. A novel approach in molecular representation learning, qcGEM integrates quantum chemistry knowledge into graph-based embeddings, offering a new dimension for AI-driven drug discovery. 2. The model

The code for DropD-SLAM is now published:

Recording From our Previous Talk given by Adam J. Broerman @adam_broerman is up! @Eng_Musinguzi_B @BrightonNuwa https://t.co/y7M5waQ6nM

Video motion and view control just became easy! Check out our new plug-and-play approach led by my brilliant students and collaborators @assaf_singer @NoamRot @mann_amir_ @RonnyKimmel @TechnionLive 🌐project page:

Achiral BODIPY, Chiral emission... Check out more details: Peptide-induced chirality transfer and circularly polarized luminescence in achiral BODIPY emitters via halogen bonding... @ChemCommun @anindita_iacs https://t.co/poa5f2ivuZ

single-cell models tend to learn from the many - and miss the rare we introduce an Adaptive Resampling approach to help models learn from underrepresented cells, improving generalization & discovery https://t.co/CIpAW1RwWA https://t.co/xfCc13dMe5 great work by @NavidiZeinab!

OUGS: Active View Selection via Object-aware Uncertainty Estimation in 3DGS Haiyi Li, Qi Chen, Denis Kalkofen, Hsiang-Ting Chen tl;dr: Gaussian parameters->covariance->diagonal Fisher Information Matrix->uncertainty https://t.co/k7lD1afVhk

Introducing Lightning Grasp, a high-performance procedural grasp synthesis algorithm that generates thousands of dexterous grasps in seconds, across diverse hands and challenging objects. ⚡️ 10-100x faster than sota. Paper: https://t.co/xscVvXzwK8 Code: https://t.co/UCLsi4J2E6

High-accuracy structure modeling for antibody-antigen complexes 1. A new computational protocol called DeepAAAssembly has been introduced to enhance the accuracy of antibody-antigen complex modeling. This method integrates deep learning-based inter-chain residue distance

Thrilled to share Entangled Schrödinger Bridge Matching (EntangledSBM), a new framework that extends classical SBM to multi-particle interacting systems where the trajectories of each particle depend on the time-evolving dynamics of neighbouring particles. ⚛️🌌 📄 Preprint:

📢ProcGen3D: Learning Neural Procedural Graphs for Image-to-3D Reconstruction @xinyi092298 learns neural procedural graphs to generate high-fidelity 3D - MCTS-guided sampling maintains consistency with the input image, even from real images! Check it out: https://t.co/RLGd2iXCwf

had a blast with @pranamanam, @AlexanderTong7, @rohitsingh8080, and xingcheng lin, kicking off the talks at the Carolina biophysics symposium. so cool to see so many talks with novel biological results enabled by ML AI/ML 🤝 bio