Topics
70 curated papers across all published weeks — sorted by disruptiveness score.
Time series reasoning is crucial to decision-making in diverse domains, including finance, energy, and scientific discovery. While existing time series foundat…
Read free explainer →
2026-W34
Recursive self-improvement (RSI) asks whether an AI system can improve the process that produces AI systems, so that the next system inherits the improvement.…
2026-W35
Anatomically plausible segmentation remains challenging because of low contrast, ambiguous boundaries, and modality-specific artifacts. Interactive segmentatio…
2026-W36
Reasoning traces from chain-of-thought models appear to offer a legible window into how a model arrives at its answer. A growing body of work treats them as su…
2026-W37
Distributed learning in embodied reinforcement-learning agents offers a degree of privacy by retaining raw sensor data on-device and transmitting only policy g…
2026-W40
LLMs are increasingly applied to cybersecurity workflows, where they are expected to translate analysts' intent into tool invocations. However, existing evalua…
2026-W41
The emergence of foundation models for molecular property prediction requires a high degree of AI data readiness, including reliable metadata annotation. Howev…
Spatial reasoning depends not only on metric properties such as distance, angle, and shape, but also on topological relations that remain invariant under conti…
2026-W38
More than 300 million people worldwide are affected by one of over 7,000 known rare diseases, yet diagnosis remains difficult because patients initially presen…
2026-W39
Medical image diagnosis has achieved significant progress with deep learning, yet existing methods often rely on isolated visual evidence and lack the ability…
Perceiving and remembering the visual world is fundamental to navigating and interacting with our environment. Current 4D foundation models, such as camera-con…
We present Swift-Image, a compact unified model for text-to-image generation, single-image editing, and multi-image editing. Our goal is to explore how far a r…
Recent research on Multimodal Sentiment Analysis (MSA) has focused on learning from language, visual, and acoustic modalities with incomplete data to infer hum…
Tool-using language model agents rely on environmental facts to decide whether an action is permitted. When a tool response omits a safety-relevant fact, an ag…
2026-W44
We introduce DeepSeek-R1, a reasoning model trained with large-scale reinforcement learning that achieves strong multi-step reasoning without extensive human-a…
2026-W30
Prefix adders are fundamental arithmetic circuits, but their design space grows exponentially with bit-width, posing significant optimization challenges. Previ…
Protein classifiers are highly sensitive to evaluation protocol: random train/test splits induce phylogenetic leakage, allowing models to exploit taxon-specifi…
Heterogeneous AI systems composed of multiple models, architectures, harnesses, or inference-time settings can improve quality and efficiency by routing querie…
We show that multimodal models possess strong reasoning abilities and that an appropriate harness can unlock their potential to solve tasks across diverse inte…
High-risk non-muscle-invasive bladder cancer (HR-NMIBC) carries substantial risks of recurrence and progression, while current clinical risk stratification rem…
Conspiratorial discourse on social media is not always expressed through explicit claims or stable lexical markers. The same surface content may express endors…
Recent advances in multimodal large language models (MLLMs) and 3D generation have expanded 3D understanding and synthesis. However, beyond visual fidelity, ta…
2026-W43
We propose Puffin-World, a unified multimodal architecture that integrates physical understanding, spatial simulation, and 3D world generation and reconstructi…
Protein modification requires navigating an immense sequence space, yet wet-lab validation remains low-throughput and costly. Although computational paradigms…
Generative visual verifiers play a crucial role in advancing multimodal intelligence. To enhance the robustness of visual verifiers in complex multimodal scena…
DeepSeek-V3 is a strong Mixture-of-Experts language model with 671B total parameters and 37B activated per token. We describe Multi-head Latent Attention, an a…
The entire ecosystem of open-source language models effectively relies on a single platform. What if this platform was forced to shut down tomorrow? Implementi…
Frontier coding agents are becoming increasingly adept at 3D modeling. Yet, they still face challenges when building objects meant to be assembled in the physi…
While data-driven 3D shape correspondence estimation has recently seen substantial progress, robust matching under partial observations and strong non-isometri…
Long-video understanding on edge devices must reason over hours of content under tight compute and bandwidth budgets. Subsampling visual tokens loses temporal…
Frontier coding agents are increasingly trusted to work autonomously for long periods, yet an agent's final response is often the only account of that work a u…
Hybrid Transformer-Mamba large language models (LLMs) enhance long-context efficiency, but their heterogeneous computation and communication patterns complicat…
Language-model agents are increasingly asked to carry out open-ended scientific research, yet their results are usually graded against a known answer, a rubric…
2026-W42
LLMs can act as scientific agents using tools to consult specialists and run experiments. Yet specialist information must enter their context, generally as tex…
Travel behavior research increasingly combines digital data collection with predictive modeling, yet these stages are often developed and evaluated separately.…
We scale sparse autoencoder interpretability techniques to frontier multimodal models, recovering monosemantic features that enable precise activation steering…
Diffusion models have shown strong potential for kilometer-scale precipitation downscaling, but their performance in geographically unseen regions and event re…
We present Llama 3, a family of foundation models that support multilinguality, coding, reasoning, and tool use. We detail pretraining data mixtures, scaling l…
We train a large-scale transformer policy on heterogeneous humanoid datasets that zero-shot adapts to new morphologies and terrains, combining imitation, reinf…
Large Language Models (LLMs) increasingly require selective removal of harmful or sensitive knowledge, called unlearning, yet existing methods and benchmarks f…
Deep neural networks are often overconfident, assigning high confidence even to incorrect predictions. Consequently, users lack a reliable signal for deciding…
We launch SenseNova-U1.5, an 8B-MoT native unified multimodal model that understands, reasons about, and generates visual content within an encoder-free and VA…
Programmable world models separate executable dynamics from visual generation, offering a promising foundation for next-generation game engines. However, their…
Naturalistic computer-use traces, passively recorded screenshots and mouse or keyboard actions, are a valuable resource for deriving symbolic, auditable, and r…
Label-free virtual staining offers a compelling, non-destructive alternative to standard histopathology; however, its clinical adoption is hindered by the comp…
Antimicrobial resistance is expected to claim 10 million lives per year by 2050, and resource-limited regions are most affected. Raman spectroscopy is a novel…
We prove a rigorous learning-theoretic separation showing that a quantum learner with access to a single clean qubit and mixed states can efficiently learn spa…
Machine-learned drive shaping improves hot-spot symmetry in inertial confinement fusion experiments, yielding a statistically significant boost in neutron yiel…
A multi-satellite fusion pipeline attributes methane plumes to facility-level sources within hours, with quantified uncertainty suitable for regulatory enforce…
Video generators and video-based world models can synthesize plausible visual trajectories, but long-horizon procedural tasks require generation to adapt to wh…
Token merging has emerged as an effective training-free strategy for accelerating diffusion models by reducing redundant token computation while minimizing gen…
Wrong automatic routing of an IT support ticket can cost more than human triage. We evaluate cost-aware abstention as an offline, cost-conditional readiness as…
An autonomous laboratory combines literature-trained language models, Bayesian optimization, and robotic synthesis to discover three previously unreported soli…
Scientific discovery increasingly requires AI systems that can reason over scientific evidence of heterogeneous modalities, interact with scientific tools and…
2026-W33
We integrate high-resolution tactile sensing into vision-language-action models, enabling contact-rich assembly tasks that pure vision policies fail, with stro…
We train neural cellular automata to design biocompatible scaffold growth programs that regenerate complex tissue geometries in silico and guide 3D bioprinting…
The fundamental goal of agentic visual reasoning is to improve the success rate of multimodal large language models (MLLMs) on complex tasks, rather than merel…
2026-W31
Recent advances in foundation models have enabled AI scientists to automate increasingly complete research workflows, from hypothesis generation and code execu…
We introduce the first Probably Approximately Correct (PAC) learning framework for general-sum concurrent stochastic games (CSGs) with transition uncertainty,…
Reinforcement learning (RL) of large language models is notoriously sensitive to small differences between training and inference engines, often referred to as…
High-fidelity 3D generation predominantly relies on scaling model capacity and data, which incurs prohibitive computational costs. This paradigm typically requ…
While Multimodal Retrieval-Augmented Generation (MM-RAG) has shown promising results, it still struggles with complex multi-hop reasoning tasks. Existing metho…
Efficiently retrieving relevant clips from large-scale driving logs is essential for data curation, model development, and safety analysis. Structured and rule…
Scaling coding agents requires a continuing supply of executable data for training, benchmarking, and continuous evaluation. Each task must couple a realistic…
Deploying autonomous computer-use agents (CUAs) locally is increasingly important for privacy, cost efficiency, and practical usability, yet improving their pe…
Transforming multimodal sources into condensed and structured media outputs can be fundamentally conceptualized as a long-horizon agentic process centered on a…
Current large language model development relies on massive, often non-permissible datasets, creating a high barrier for researchers committed to open-source an…
AI agents are increasingly used for programming, but do not provide any guarantee on the correctness of generated code. Verified code generation, in which an a…
Tabular-to-image methods have emerged as novel approaches to leverage the high predictive performance of convolutional neural networks and vision transformers.…
Snapshot Compressive Imaging (SCI) offers an efficient solution for high-speed video acquisition and, under exposure-time camera--scene relative motion, multi-…
