research

The Anatomy of the CTC Oracle Gap: Acoustic Exhaustion and Linguistic Recovery

arXiv:2606.23306v1 Announce Type: cross Abstract: We study the limits of CTC-internal scoring for N-best hypothesis selection and locate the information bottleneck separating acoustic confidence from

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

The Cost Geometry of Belief: finite-resource inference under noisy observation

DGX agent

arXiv:2606.21585v1 Announce Type: new Abstract: We equip the space of beliefs with a cost geometry (what it costs to pass from one belief to another): optimal transport in Wasserstein space, reweighte

researcharxiv-cs-lg

23 Jun 2026

The Download: the future of chipmaking and Anthropic’s government clash

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. The $400 million machine powering the future of chipmaking It’

researchmit-tech-review

The Energy Consumption of Transformer Fine-Tuning: A Roofline-Inspired Scaling Model

arXiv:2606.23546v1 Announce Type: new Abstract: Transformer-based models underpin modern natural language processing but incur rapidly growing computational and energy costs. As training scales in bot

The First Assessment of PhiSat-2 Imagery for Monocular Building Height Estimation

arXiv:2603.29245v3 Announce Type: replace Abstract: Monocular building height estimation from optical imagery is important for characterizing urban vertical structure, yet remains challenging due to t

The Keras team is gonna be at the AI Systems DevLabs in Sunnvale tomorrow. Come say hi if you're around!

researchfrancois-chollet--x

François Chollet announced that members of the Keras team would be attending the AI Systems DevLabs event in Sunnyvale and invited people to meet them there. This post indicates a scheduled appearance

The Optimal Token Baseline: Variance Reduction for Long-Horizon LLM-RL

arXiv:2602.07078v2 Announce Type: replace Abstract: Reinforcement Learning (RL) for Large Language Models (LLMs) often suffers from training collapse in long-horizon tasks due to exploding gradient va

The Optimization Landscape of Caratheodory Decomposition of Toeplitz Covariances

arXiv:2511.01605v2 Announce Type: replace Abstract: Toeplitz covariance estimation is a classical problem in statistical signal processing, yet the geometry of the Gaussian maximum-likelihood objectiv

The Trilemma of Truth in Large Language Models

arXiv:2506.23921v5 Announce Type: replace-cross Abstract: The public often attributes human-like qualities to large language models (LLMs), assuming that they 'know' certain things. In reality, LLMs e

The Unreasonable Effectiveness of VLMs for Zero-shot Procedural Mistake Detection

arXiv:2606.21579v1 Announce Type: new Abstract: Procedural mistake detection is important for quality control and user assistance across many disciplines. Recent work in this field has achieved signif

ThermoLLM: Thermodynamics-Aware HVAC Control with Spatial-Semantic Knowledge Graph

arXiv:2606.22911v1 Announce Type: cross Abstract: Multi-zone HVAC control is a spatial decision problem in which indoor thermal evolution and control decisions depend not only on outdoor conditions an

TIDY: Thermal Infrared Image Denoising via Wavelet Domain Entropy and Directional Stripe Index

arXiv:2606.19813v1 Announce Type: cross Abstract: Thermal infrared (TIR) imaging has been a popular choice for field robotics due to its robust perception capability under low light visual degradation

TooBad: Backdoor Diffusion Models with Ultra-Low Poison Rate and Imperceptible Trigger

arXiv:2606.23362v1 Announce Type: cross Abstract: Diffusion models (DMs), despite their impressive capabilities across a wide range of generative tasks, have been shown to be vulnerable to backdoor at

Towards a Novel Wearable Robotic Vest for Hemorrhage Suppression

arXiv:2602.01448v2 Announce Type: replace Abstract: This paper introduces a novel robotic system designed to manage severe bleeding in emergency scenarios, including unique environments like space sta

researcharxiv-cs-ro

Towards Accurate and Robust Surveillance Roadside IVD via Trackletized Audio-Visual Reasoning

arXiv:2606.22299v1 Announce Type: new Abstract: Idling Vehicle Detection (IVD) seeks to determine, at the final frame of a video clip, whether any vehicle is idling, meaning the vehicle is stationary

Towards CSI-Native Foundation Models: A Channel-Adaptive Roadmap for 6G

arXiv:2606.20670v1 Announce Type: new Abstract: Wireless foundation models offer a path toward reusable channel state information (CSI) intelligence for sixth-generation (6G) systems. However, existin

Towards Robust Training in NNGPT AutoML Pipeline: A Loss-Optimizer Pairing Selection Study

arXiv:2606.20933v1 Announce Type: new Abstract: The choice of loss function and optimizer is an important decision, that shapes further model training. Yet automated architecture search pipelines (Aut

Towards Understanding the Power and Limits of the Muon Optimizer: A River-Valley Perspective

arXiv:2606.21514v1 Announce Type: new Abstract: Recently, Muon has gained substantial attention as an appealing alternative to Adam-like optimizers, with many works highlighting its advantages through

TraceMark-LDM: Authenticatable Watermarking for Latent Diffusion Models via Binary-Guided Rearrangement

arXiv:2503.23332v2 Announce Type: replace Abstract: Image generation algorithms are increasingly integral to diverse aspects of human society, driven by their practical applications. However, insuffic

TriMotion: Modality-Agnostic Camera Control for Video Generation

arXiv:2606.20774v1 Announce Type: new Abstract: Camera motion control is essential for directing viewpoint changes in generative systems. However, existing methods typically condition the generation p

Trustworthy MRI Reconstruction via Bayesian Uncertainty Quantification with Sparsity Prior Models

arXiv:2606.17343v2 Announce Type: replace Abstract: We propose a novel Bayesian framework for joint image reconstruction and uncertainty quantification from compressed sensing magnetic resonance imagi

Turning Tabular Foundation Models into Graph Foundation Models

arXiv:2508.20906v3 Announce Type: replace Abstract: While foundation models have revolutionized fields such as natural language processing and computer vision, their potential in graph machine learnin

Two Layers of Instability in Causal Estimation

arXiv:2606.21185v1 Announce Type: cross Abstract: There is a precise sense in which drawing causal inferences from observational data is hard, even when identifiability is assumed. In particular, Robi

Ultra-Peripheral Collisions as a Nuclear-Structure Interferometer with Interpretable Multitask Deep Learning

arXiv:2606.23353v1 Announce Type: cross Abstract: Precise knowledge of nuclear structure is essential across fundamental physics, yet probing these structures is notoriously difficult. To address this

Ultrasound imaging turns a robot hand into a skillful mimic

Our hands are the nimblest parts of our bodies, coordinating 34 muscles, 27 joints, and over 100 tendons and ligaments to perform countless nuanced movements and gestures. So far, robots have been not

researchmit-tech-review

Uncertainty-Aware Domain Adaptation for Vitiligo Segmentation in Clinical Photographs

arXiv:2512.11791v2 Announce Type: replace Abstract: Accurately quantifying vitiligo extent in routine clinical photographs is crucial for longitudinal monitoring of treatment response. We propose a tr

Understanding Latent Flow Models for Tabular Data Synthesis: Targets, Paths, and Sampling

arXiv:2606.20878v1 Announce Type: new Abstract: Synthetic tabular data enables microdata sharing in regulated domains, yet deploying continuous-time generative models requires balancing analytical uti

Unified Motion-Action Modeling for Heterogeneous Robot Learning

arXiv:2606.16917v3 Announce Type: replace Abstract: We present Unified Motion-Action (UMA) Model, an approach that uses 3D object motion trajectories as a shared interface to bridge visuomotor control

researcharxiv-cs-ro

UniFS: Unified Fast-to-Slow Hierarchical Architecture for Vision-Language-Action Models

arXiv:2606.22794v1 Announce Type: new Abstract: Mainstream Fast-Slow dual system vision-language-action models decouple a high-frequency action expert from a low-frequency vision-language model for ef

researcharxiv-cs-ro

UniSLAD: A Unified Framework for Structural and Logical Industrial Visual Anomaly Detection

arXiv:2606.20768v1 Announce Type: new Abstract: Visual anomaly detection is a fundamental task in industrial automation. While existing approaches have achieved notable progress in identifying structu

Universal Encoders for Modular Relational Deep Learning

arXiv:2606.21434v1 Announce Type: new Abstract: Relational Deep Learning (RDL) models multi-tabular databases as temporal heterogeneous graphs for end-to-end representation learning. While RDL is evol

Universal priors: solving empirical Bayes via Bayesian inference and pretraining

arXiv:2602.15136v2 Announce Type: replace-cross Abstract: We theoretically justify the recent empirical finding of [Teh et al., 2025] that a transformer pretrained on synthetically generated data achi

UniverSat: Resolution- and Modality-Agnostic Transformers for Earth Observation

arXiv:2606.23503v1 Announce Type: new Abstract: Vision Transformers (ViT) dominate computer vision. However, their reliance on rigid patch projectors hinders transfer to Earth Observation (EO), where

Unlocking In-Context Learning in Audio-Language Models from Decentralized Medical Audio

arXiv:2606.23243v1 Announce Type: new Abstract: Clinical audio diagnosis in low-resource settings requires models that identify conditions from minimal examples without large annotated corpora. We pro

Unmasking LAION-5B: Age, Gender, Race, and Emotion Biases in Large-Scale Image Datasets

arXiv:2606.23204v1 Announce Type: new Abstract: Large-scale image-text datasets, such as LAION-5B, are foundational to modern AI systems, yet their vast scale and uncurated nature raise significant co

Unsupervised Disentanglement Without Compromises : How Functional Orthogonality Enforces Identifiability

arXiv:2606.21385v1 Announce Type: new Abstract: This paper explores unsupervised disentangled representation learning from a functional perspective. We define latent concepts as factors that influence

Unsupervised Susceptibility Distortion Correction of EPI without Calibration Scans via Image Translation-Based Registration

arXiv:2606.21588v1 Announce Type: cross Abstract: Functional magnetic resonance imaging (fMRI) utilizes echo-planar imaging (EPI) to capture blood-oxygen-level-dependent (BOLD) signals with high tempo

Urban Power Grid Topology and Hierarchy Identification from Open Data

arXiv:2606.21352v1 Announce Type: new Abstract: Understanding the complex topology and hierarchy of urban power grid is crucial for energy prognosis, power flow management, and system resilience analy

Variance-Tilted Diffusion Models for Diverse Sampling

arXiv:2606.22239v1 Announce Type: cross Abstract: Diffusion models are typically sampled independently, even when the downstream objective is to obtain a diverse set of candidates. We introduce a vari

Venice-H1: Failure-Aware Query Re-Ranking with Multi-Scale Grid Signatures for Referring Image Segmentation

arXiv:2606.22546v1 Announce Type: new Abstract: Modern Referring Image Segmentation (RIS) systems generate multiple candidate masks per expression but rely on a simple heuristic--typically the argmax

VeriBound: PAC-Bayesian Generalization Bounds for Process Reward Models Trained with Formal Verification Tools

arXiv:2606.20740v1 Announce Type: cross Abstract: Process Reward Models (PRMs) provide step-level verification for Large Language Model (LLM) reasoning, yet their training data acquisition remains a b

Video2Code: Generating Interactive Webpages from UI Videos via Action-Aware Revisit

arXiv:2606.20711v1 Announce Type: new Abstract: UI videos provide a natural input for generating interactive webpages, as they capture both webpage appearance and action-triggered state transitions. H

WarPGNN: A Parametric Thermal Warpage Analysis Framework with Physics-aware Graph Neural Network

arXiv:2603.18581v2 Announce Type: replace-cross Abstract: With the advent of system-in-package (SiP) chiplet-based design and heterogeneous 2.5D/3D integration, thermal-induced warpage has become a cr

When Calibration Fails the Vulnerable Hospital: Federated Conformal Risk Control via Risk-Curve Shrinkage

arXiv:2606.20115v2 Announce Type: replace-cross Abstract: Conformal risk control (CRC) provides distribution-free guarantees on segmentation quality by calibrating a prediction-set threshold on held-o

When Does a Video-Language Model Stop Watching? Reward Strength Controls the Formation and Reversal of Visual Shortcuts in Multimodal RLVR

arXiv:2606.22043v1 Announce Type: cross Abstract: Reinforcement learning with verifiable rewards (RLVR) is increasingly applied to large vision-language models (LVLMs), yet outcome-only optimization c

When Is an LLM Worth It for Hyperparameter Optimization? A Budget-Matched Study on Tabular Data Finds the Warm-Start Is a Default Configuration, Not the Model

arXiv:2606.21641v1 Announce Type: new Abstract: Large language models (LLMs) have been proposed as hyperparameter-optimization (HPO) advisors that 'warm-start' search from prior knowledge, proposing s

World Action Models: A Survey

arXiv:2606.20781v1 Announce Type: cross Abstract: World Action Models (WAMs) are embodied predictive-action models that make a forecast of the future available to action. Recent WAMs repurpose large v

XmoPipe: A Pipeline for Large-Scale In-the-Wild Human Motion Dataset Construction

arXiv:2606.20731v1 Announce Type: new Abstract: Large-scale human motion datasets are essential for training robust motion models for analysis, synthesis, and understanding. While marker-based motion