AI & Machine Learning

An online length-aware scheduling strategy that eliminates training 'bubbles' during the rollout phase of LLM reinforcement learning.

A bilevel framework where an outer LLM loop meta-optimizes an inner autoresearch loop by autonomously generating and injecting Python code at runtime.

Integrates tactile perception into video-action models to enable high-fidelity force modulation in contact-rich robotic tasks.

Enables training of monocular novel-view synthesis models using entirely unpaired, in-the-wild internet images.

Leverages human gaze tracking to assign non-uniform token density in diffusion models, creating perceptually perfect images with significantly less compute.

Replaces visual token compression with sparse, dynamically selected vision-language interactions in VLLMs.

A unified reinforcement learning framework that jointly optimizes reasoning (text) and synthesis (image) for interleaved multimodal generation.

Introduces on-the-fly quantization that calibrates to individual prompts during inference, solving the 'domain shift' problem where standard quantization fails on unseen data.

Provides a statistically rigorous framework to evaluate model performance and reliability after cherry-picking or selecting models based on the same test data.

Develops a differentially private RLHF pipeline that decouples private reward learning from policy optimization, achieving strong alignment on Gemma-2B-IT with privacy guarantees.

AI is actually the most confident when it's completely making stuff up.

Future phones might have 'liquid' antennas that literally swim around inside the device to hunt down a better signal.

A massive study found women do way more innovative science than men, but they still get robbed when it's time for the credit.

Scientists found a way to make a basic home computer screw up math exactly like a super-expensive AI chip does.

A core rule of tech just got an update, and it turns out those fancy AI chips might eventually be totally useless.

New 360-degree video treats things on screen like they have gravity, just so it can predict exactly where you're gonna look next.

Your future phone might have antennas that physically slide along tracks to 'pinch' the best Wi-Fi signal possible.

An AI just 'figured out' how to lock down its own code using high-level math without a human ever telling it how.

Engineers built 'invisible' backdoors into computer chips that are so well-hidden, even the most powerful microscopes can't find them.

Scientists found one single math formula that explains why everything from stock market crashes to earthquakes actually happens.

Researchers built an AI sensor that 'thinks' using light ripples, letting it spot objects in about 25 billionths of a second.

Researchers found one 'master' math trick that can recreate every single function on your old scientific calculator.

There’s a new AI that can tell you an animal’s whole lifestyle and what it looks like just by listening to it make a sound.

A new voting system lets you check if a national election was legit using just basic math and zero computers.

New math can spot life-threatening internal bleeding in patients before doctors can even see it.

Those single scores we use to rank people on things like intelligence might actually be mathematical illusions.

AI can now map out the secret relationships between terrorist groups that they try to keep hidden.

Achieves over 10x faster sampling for diffusion language models by shifting the process into continuous semantic space.

Integrates fast scalar rewards with slow generative CoT reasoning to reduce reward model token consumption by 20%.

Enables precise prompt routing by predicting the expected reward of a model before any response is generated.

Introduces a training strategy where Transformers 'think' in latent space before committing to discrete tokens.

Composes pre-trained unimanual robotic policies into complex bimanual tasks without requiring bimanual demonstration data.

Sets a new state-of-the-art for intracortical speech decoding with 14.3% phoneme error rate using a multitask Transformer.

Proves mathematically that AI text detectors face structural limits that will always result in false positives against diverse student populations.

The first foundation model for zero-shot prediction of joint probability distributions in coupled time series.

Reduces Tree of Thought (ToT) computational overhead by up to 75% using plug-and-play predictors for pruning.

Formalizes 'Introspection' in LLMs and proves they have privileged access to their own policy logic beyond mere self-simulation.

Releases an offline search-and-browse pipeline with 97K long-horizon trajectories for training 'Deep Research' agents.

Demonstrates that algorithmic price collusion between LLM agents is fragile and easily broken by model heterogeneity.

STAC achieves a 10x memory reduction and 4x speedup for real-time streaming 3D reconstruction using spatio-temporal cache compression.

AgentComm-Bench is the first benchmark to stress-test cooperative embodied AI under realistic wireless impairments like packet loss and bandwidth collapse.

InjectFlow is a training-free method that fixes semantic degradation and bias in Flow Matching models by injecting orthogonal semantics into the velocity field.

DiffMark enables multi-bit watermarking that is transferable across different frozen diffusion models with a 45x speedup over current methods.

Reason-to-Transmit introduces deliberative communication for multi-agent systems, where agents reason about *why* a message benefits the receiver rather than just broadcasting features.

BubbleRAG enables high-precision retrieval-augmented generation over black-box Knowledge Graphs where the schema and structure are unknown.

VGS-Decoding is a training-free method to mitigate medical VLM hallucinations by reweighting token probabilities based on their visual dependency.

This paper demonstrates that Model Context Protocol (MCP) can outperform traditional RAG for quantitative financial Q&A by interacting directly with structured data APIs.

Researchers identify a 'selection bottleneck' that mathematically determines when diverse agent teams outperform homogeneous self-consistency teams.

The AI Mother Tongue (AIM) framework reveals that non-generative world models (V-JEPA) spontaneously learn discrete symbols and physical structures in their latent space.

GEM is the first native graph-based index for multi-vector (ColBERT-style) retrieval, achieving up to 16x speedups over existing single-vector index adaptations.