AI Infrastructure

Integrate Physical AI Capabilities into Existing Apps with NVIDIA Omniverse Libraries

Physical AI—AI systems that perceive, reason, and act in physically grounded simulated environments—is changing how teams design and vali...

UMKC Announces New Master of Science in Artificial Intelligence

UMKC announces a new Master of Science in Artificial Intelligence program aimed at addressing workforce demand for AI expertise, set to l...

Dell and HIVE partner to deploy Nvidia’s next-generation AI chips

[2510.18478] Safe But Not Sorry: Reducing Over-Conservatism in Safety Critics via Uncertainty-Aware Modulation

This article presents the Uncertain Safety Critic (USC), a novel approach to enhance safety in reinforcement learning (RL) by balancing s...

[2509.15194] Evolving Language Models without Labels: Majority Drives Selection, Novelty Promotes Variation

The paper presents EVOL-RL, a novel framework for evolving language models without labels, balancing majority-driven stability with novel...

[2508.12907] SNAP-UQ: Self-supervised Next-Activation Prediction for Single-Pass Uncertainty in TinyML

The paper presents SNAP-UQ, a novel method for single-pass uncertainty estimation in TinyML, enhancing reliability in on-device monitorin...

[2501.16534] Targeting Alignment: Extracting Safety Classifiers of Aligned LLMs

This article presents a novel technique for extracting safety classifiers from aligned large language models (LLMs) to address vulnerabil...

[2501.14406] Adaptive Rank Allocation for Federated Parameter-Efficient Fine-Tuning of Language Models

The paper presents FedARA, an innovative framework for federated parameter-efficient fine-tuning of language models, addressing data hete...

[2508.10480] Pinet: Optimizing hard-constrained neural networks with orthogonal projection layers

The paper introduces $ ext{Pinet}$, a novel output layer for neural networks that optimizes hard constraints using orthogonal projection ...

[2507.12257] Robust Causal Discovery in Real-World Time Series with Power-Laws

This paper presents a novel method for causal discovery in time series data, leveraging power-law distributions to enhance robustness aga...

[2506.14202] DiffusionBlocks: Block-wise Neural Network Training via Diffusion Interpretation

The paper introduces DiffusionBlocks, a framework for block-wise training of neural networks that reduces memory bottlenecks while mainta...

[2602.11348] AgentNoiseBench: Benchmarking Robustness of Tool-Using LLM Agents Under Noisy Condition

The paper introduces AgentNoiseBench, a framework for evaluating the robustness of tool-using LLM agents under noisy conditions, highligh...

[2505.19427] WINA: Weight Informed Neuron Activation for Accelerating Large Language Model Inference

The paper introduces WINA, a novel framework for efficient inference in large language models (LLMs) that optimally combines hidden state...

[2602.02050] Rethinking the Role of Entropy in Optimizing Tool-Use Behaviors for Large Language Model Agents

This article explores the role of entropy in optimizing tool-use behaviors for Large Language Model (LLM) agents, highlighting the correl...

[2504.06768] FedMerge: Federated Personalization via Model Merging

The paper introduces FedMerge, a novel approach in federated learning that enables personalized model creation for clients by merging mul...

[2504.05615] FedEFC: Federated Learning Using Enhanced Forward Correction Against Noisy Labels

The paper presents FedEFC, a novel approach to federated learning that addresses the challenges posed by noisy labels through techniques ...

[2601.01569] CaveAgent: Transforming LLMs into Stateful Runtime Operators

CaveAgent introduces a novel framework that transforms LLMs into stateful runtime operators, enhancing their ability to manage complex ta...

[2510.12121] Precise Attribute Intensity Control in Large Language Models via Targeted Representation Editing

This paper introduces a method for precise control of attribute intensities in Large Language Models (LLMs) through targeted representati...

[2502.07274] Forget Forgetting: Continual Learning in a World of Abundant Memory

The paper explores continual learning (CL) in AI, proposing a shift from minimizing memory usage to leveraging abundant memory while addr...

[2502.00213] Understanding Transformer Optimization via Gradient Heterogeneity

This paper explores the optimization challenges of Transformer models, focusing on gradient heterogeneity and its impact on convergence w...

[2509.00074] Language and Experience: A Computational Model of Social Learning in Complex Tasks

This article presents a computational model that explores how humans and AI can integrate linguistic guidance and direct experience for e...

[2409.04332] Amortized Bayesian Workflow

The paper presents an Amortized Bayesian Workflow that combines fast amortized inference with accurate MCMC techniques, optimizing Bayesi...

[2411.04760] Zero-Shot Temporal Resolution Domain Adaptation for Spiking Neural Networks

This paper presents novel domain adaptation methods for Spiking Neural Networks (SNNs) to address performance drops due to mismatched tem...