papers checked against Recognition

60 of 33923 papers

• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks

  cs.LG · 2015-11-19 · 3 theorem links

  DCGANs with architectural constraints learn a hierarchy of representations from object parts to scenes in both generator and discriminator across image datasets.

• LSUN: Construction of a Large-scale Image Dataset using Deep Learning with Humans in the Loop

  cs.CV · 2015-06-10 · 3 theorem links

  LSUN dataset of one million images per category across 30 classes is constructed via iterative human-in-the-loop deep learning labeling.

• Unleashing Large-Scale Video Generative Pre-training for Visual Robot Manipulation

  cs.RO · 2023-12-20 · 3 theorem links

  A GPT-style model pre-trained on large video datasets achieves 94.9% success on CALVIN multi-task manipulation and 85.4% zero-shot generalization, outperforming prior baselines.

• Improving Video Generation with Human Feedback

  cs.CV · 2025-01-23 · 3 theorem links

  A human preference dataset and VideoReward model enable Flow-DPO and Flow-NRG to produce smoother, better-aligned videos from text prompts in flow-based generators.

• MixGRPO: Unlocking Flow-based GRPO Efficiency with Mixed ODE-SDE

  cs.AI · 2025-07-29 · 3 theorem links

  MixGRPO speeds up GRPO for flow-based image generators by restricting SDE sampling and optimization to a sliding window while using ODE elsewhere, cutting training time by up to 71% with better alignment performance.

• C-Pack: Packed Resources For General Chinese Embeddings

  cs.CL · 2023-09-14 · 3 theorem links

  C-Pack releases a new Chinese embedding benchmark, large training dataset, and optimized models that outperform priors by up to 10% on C-MTEB while also delivering English SOTA results.

• Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models

  cs.LG · 2024-03-28 · 3 theorem links

  Sparse feature circuits are introduced as interpretable causal subnetworks in language models, supporting unsupervised discovery of thousands of circuits and a method called SHIFT to improve classifier generalization by ablating irrelevant features.

• Why Language Models Hallucinate

  cs.CL · 2025-09-04 · 3 theorem links

  Language models hallucinate due to statistical pressures from training pipelines that treat uncertainty as binary classification errors and from evaluations that penalize admitting ignorance.

• AndroidWorld: A Dynamic Benchmarking Environment for Autonomous Agents

  cs.AI · 2024-05-23 · 3 theorem links

  AndroidWorld is a dynamic, reproducible Android benchmark that generates unlimited natural-language tasks for autonomous agents and shows current agents succeed on only 30.6 percent of them.

• An Overview of Multi-Task Learning in Deep Neural Networks

  cs.LG · 2017-06-15 · 3 theorem links

  Multi-task learning succeeds in deep networks by sharing parameters across related tasks and selecting helpful auxiliary tasks to improve generalization and efficiency.

• Chronos: Learning the Language of Time Series

  cs.LG · 2024-03-12 · 3 theorem links

  Chronos pretrains transformer models on tokenized time series to deliver strong zero-shot forecasting across diverse domains.

• Reinforced Self-Training (ReST) for Language Modeling

  cs.CL · 2023-08-17 · 3 theorem links

  ReST improves LLM translation quality on benchmarks via offline RL on self-generated data, achieving gains in a compute-efficient way compared to typical RLHF.

• Seedream 3.0 Technical Report

  cs.CV · 2025-04-15 · 3 theorem links

  Seedream 3.0 improves bilingual image generation through doubled defect-aware data, mixed-resolution training, cross-modality RoPE, representation alignment, aesthetic SFT, VLM reward modeling, and importance-aware timestep sampling for 4-8x faster inference at up to 2K resolution.

• Structural Interpretations of Protein Language Model Representations via Differentiable Graph Partitioning

  cs.LG · 2026-05-09 · 3 theorem links

  SoftBlobGIN combines ESM-2 representations with protein contact graphs via a lightweight GNN and differentiable substructure pooling to achieve 92.8% accuracy on enzyme classification, raise binding-site AUROC to 0.983, and generate auditable structural explanations without retraining the language模型

• SDG-MoE: Signed Debate Graph Mixture-of-Experts

  cs.LG · 2026-05-08 · 3 theorem links

  SDG-MoE introduces learned signed interaction graphs and disagreement-gated deliberation among experts in MoE architectures, yielding 19.8% better validation perplexity than the strongest baseline.

• ABRA: Agent Benchmark for Radiology Applications

  cs.CV · 2026-05-11 · 3 theorem links

  ABRA shows radiology agents excel at tool execution (89%+) but struggle with outcomes (0-25%), with oracle perception raising outcomes to 69-100%, identifying perception as the primary bottleneck.

• A New Kind of Network? Review and Reference Implementation of Neural Cellular Automata

  cs.CV · 2026-04-27 · 3 theorem links

  A review of Neural Cellular Automata that unifies existing approaches with a modular framework, notation, and reference code in NCAtorch.

• Advanced Virgo: a 2nd generation interferometric gravitational wave detector

  gr-qc · 2014-08-18 · 3 theorem links

  Advanced Virgo upgrades the Virgo detector to increase the number of observable galaxies for gravitational wave detection by three orders of magnitude.

• CosyVoice 2: Scalable Streaming Speech Synthesis with Large Language Models

  cs.SD · 2024-12-13 · 3 theorem links

  CosyVoice 2 delivers human-parity naturalness and near-lossless streaming speech synthesis by combining finite-scalar quantization, a streamlined pre-trained LLM, and chunk-aware causal flow matching on large multilingual data.

• Proteus: A Self-Evolving Red Team for Agent Skill Ecosystems

  cs.CR · 2026-05-12 · 3 theorem links

  Proteus demonstrates that adaptive red-teaming achieves 40-90% attack success after five rounds and bypasses even strong auditors at up to 41% joint success, revealing that static skill vetting underestimates residual risk.

• Condensed Mathematics and Complex Geometry

  math.CV · 2026-05-12 · 3 theorem links

  Lecture notes use condensed mathematics to reprove finiteness of coherent cohomology, Serre duality, GAGA, and Hirzebruch-Riemann-Roch for compact complex manifolds.

• Intrinsic volumes of $\ell_p$-balls and a continuum of Maxwell--Poincar\'e--Borel laws for their curvature measures

  math.PR · 2026-05-12 · 3 theorem links

  Explicit formulas for intrinsic volumes of ℓ_p-balls via one-dimensional integrals with special function F_p, plus Maxwell-Poincaré-Borel limit laws for curvature measures in high dimensions.

• DeBERTa: Decoding-enhanced BERT with Disentangled Attention

  cs.CL · 2020-06-05 · 3 theorem links

  DeBERTa improves BERT-style models by separating content and relative position in attention and adding absolute positions to the decoder, yielding consistent gains on NLU and NLG tasks and the first single-model superhuman score on SuperGLUE.

• Benchmarking and Resource Analysis for Augmented-Lagrangian Quantum Hamiltonian Descent

  quant-ph · 2026-05-12 · 3 theorem links

  AL-QHD benchmarks on nonconvex test functions and ACOPF power problems show useful accuracy at fixed qubit cost but require roughly 10^8 T gates for realistic instances.

• Weil-Moore anima

  math.NT · 2026-05-12 · 3 theorem links

  The Weil-Moore anima refines the Weil group into a space with higher homotopy groups to improve its cohomological behavior for number fields.

• Chaos and epoch structure in the deformed Mixmaster universe

  gr-qc · 2026-05-12 · 3 theorem links

  GUP deformation shortens Kasner epochs in the Mixmaster universe while polymerization lengthens them, with combined effects adding linearly and chaos strength depending on the deformation parameters.

• Discrete Diffusion Modeling by Estimating the Ratios of the Data Distribution

  stat.ML · 2023-10-25 · 3 theorem links

  Score entropy loss enables discrete diffusion models (SEDD) that cut perplexity 25-75% versus prior diffusion methods and outperform GPT-2 on language modeling while supporting infilling and compute-quality tradeoffs.

• String Diagrams for Quantum Foundations, Computing and Natural Language Processing

  quant-ph · 2026-05-12 · 3 theorem links

  String diagrams formalize constructor theory with locality-composition conflicts, enable wave-based Boolean logic design and optimization, and map Urdu text circuits equivalently to English ones up to gate translation in DisCoCirc.

• Theta functions for singular curves

  math.AG · 2026-05-11 · 3 theorem links

  A theta function is built on the compactified generalized Jacobian of singular curves to produce universal sections of line bundles of given degree via the translation action and Abel map restriction.

• A Barrier-Metric First-Order Method for Linearly Constrained Bilevel Optimization

  math.OC · 2026-05-12 · 3 theorem links

  A barrier-smoothed first-order method achieves stationarity rates of tilde O(K to the -2/3) deterministic and tilde O(K to the -2/5) stochastic for linearly constrained bilevel optimization.

• The Geometry of Spectral Fluctuations: On Near-Optimal Conditions for Universal Gaussian CLTs, with Statistical Applications

  math.ST · 2026-05-12 · 3 theorem links

  The GHOST framework gives near-optimal conditions for universal Gaussian CLTs on linear spectral statistics of sample covariance matrices, with explicit mean-covariance corrections from a bilinear fourth-order kernel and applications to corrected sphericity tests.

• PRISM: : Planning and Reasoning with Intent in Simulated Embodied Environments

  cs.RO · 2026-05-12 · 3 theorem links

  PRISM is a tiered benchmark with 300 human-verified tasks across five photorealistic apartments that diagnoses embodied agent failures in basic ability, reasoning ability, and long-horizon ability using an agent-agnostic API.

• Native Explainability for Bayesian Confidence Propagation Neural Networks: A Framework for Trusted Brain-Like AI

  cs.AI · 2026-05-12 · 3 theorem links

  BCPNNs are inherently interpretable models, and this work supplies a taxonomy, sixteen explanation primitives, and configuration artifacts to make their decisions auditable without post-hoc tools.

• On $2$-factors of Hamiltonian graphs

  math.CO · 2026-05-11 · 3 theorem links

  Large Hamiltonian graphs with minimum degree n to the power 1 minus a small epsilon contain a 2-factor consisting of exactly k cycles.

• From System 1 to System 2: A Survey of Reasoning Large Language Models

  cs.AI · 2025-02-24 · 3 theorem links

  The survey organizes the shift of LLMs toward deliberate System 2 reasoning, covering model construction techniques, performance on math and coding benchmarks, and future research directions.

• Spatial overhead reduction for 2D hypergraph product codes

  quant-ph · 2026-05-11 · 3 theorem links

  A qubit-reduction method for hypergraph product codes preserves dimension, distance, and fault-tolerance properties, producing smaller codes such as [[441,64,6]] from [[610,64,6]] with comparable noise performance and compatibility with logical gates.

• Context-Gated Associative Retrieval: From Theory to Transformers

  cond-mat.dis-nn · 2026-05-08 · 3 theorem links

  Context gating in associative memories boosts inter-memory separation and sparsity for exponential retrieval gains, admits a unique fixed point driven by direct bias and feedback, and matches in-context learning dynamics in transformers like Llama-3.

• Nautilus: From One Prompt to Plug-and-Play Robot Learning

  cs.RO · 2026-05-12 · 3 theorem links

  NAUTILUS is a prompt-driven harness that automates plug-and-play adapters, typed contracts, and validation for policies, benchmarks, and robots in learning research.

• Size Extensive Auxiliary-Field Quantum Monte Carlo with Perturbative Coupled Cluster Trial Wavefunction

  physics.chem-ph · 2026-05-11 · 3 theorem links

  A perturbative CCSD trial wavefunction renders AFQMC size-extensive with negligible bias, matching CISD-level accuracy on small systems while avoiding infrared divergence in the uniform electron gas thermodynamic limit unlike CCSD(T).

• Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolation

  cs.CL · 2021-08-27 · 3 theorem links

  ALiBi enables transformers trained on length-1024 sequences to extrapolate to length-2048 with the same perplexity as a sinusoidal model trained on 2048, while training 11% faster and using 11% less memory.

• The 2-part of the Bloch-Kato conjecture, and indivisibility results, for $K_2$ of some elliptic curves

  math.NT · 2026-05-11 · 3 theorem links

  Authors prove 2-indivisibility of a constructed K2 element on elliptic curves E_u despite explicit 2-divisibility in a related kernel, and numerically support the 2-part of Bloch-Kato for L(E_u,2).

• Diffusion Posterior Sampling for General Noisy Inverse Problems

  stat.ML · 2022-09-29 · 3 theorem links

  Diffusion models solve noisy (non)linear inverse problems via approximated posterior sampling that blends diffusion steps with manifold gradients without strict consistency projection.

• FEFormer: Frequency-enhanced Vision Transformer for Generic Knowledge Extraction and Adaptive Feature Fusion in Volumetric Medical Image Segmentation

  eess.IV · 2026-05-12 · 3 theorem links

  A frequency-enhanced Vision Transformer with FDSA, FGMLP, WAFF, and FCSB modules delivers superior volumetric medical image segmentation performance and efficiency over prior state-of-the-art methods.

• Anisotropic Cosmology with interacting Dark Energy in f(R,T) Gravity: A Data-Constrained & independent Approach

  gr-qc · 2026-05-09 · 3 theorem links

  An f(R,T) gravity model in anisotropic spacetime is reconstructed via a variable deceleration parameter and tightly constrained by Hubble and Pantheon+ data, reproducing the deceleration-to-acceleration transition and approaching de Sitter expansion.

• Show Your Work: Scratchpads for Intermediate Computation with Language Models

  cs.LG · 2021-11-30 · 3 theorem links

  Training language models to generate intermediate computation steps on a scratchpad enables them to perform multi-step tasks such as long addition and arbitrary program execution that they otherwise fail at.

• BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension

  cs.CL · 2019-10-29 · 3 theorem links

  BART introduces a denoising pretraining method for seq2seq models that matches RoBERTa on GLUE and SQuAD while setting new state-of-the-art results on abstractive summarization, dialogue, and QA with up to 6 ROUGE gains.

• Relational inductive biases, deep learning, and graph networks

  cs.LG · 2018-06-04 · 3 theorem links

  Graph networks unify graph-based neural methods into a general framework with strong relational inductive biases to support combinatorial generalization and structured reasoning in AI.

• World Simulation with Video Foundation Models for Physical AI

  cs.CV · 2025-10-28 · 3 theorem links

  Cosmos-Predict2.5 unifies text-to-world, image-to-world, and video-to-world generation in one model trained on 200M clips with RL post-training, delivering improved quality and control for physical AI.

• G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment

  cs.CL · 2023-03-29 · 3 theorem links

  G-Eval uses GPT-4 with chain-of-thought and form-filling to reach 0.514 Spearman correlation with humans on summarization, beating prior NLG metrics while noting a bias toward LLM outputs.

• Assessing requirements to scale to practical quantum advantage

  quant-ph · 2022-11-14 · 3 theorem links

  A layered resource estimation framework applied to three quantum applications shows practical advantage requires 10^5-10^6 physical qubits, driven by size, speed, and controllability.

• Open-Reasoner-Zero: An Open Source Approach to Scaling Up Reinforcement Learning on the Base Model

  cs.LG · 2025-03-31 · 3 theorem links

  A simple PPO-based RL training pipeline on base models scales reasoning performance and response length, outperforming prior work on math and science benchmarks with one-tenth the training steps.

• Microsoft COCO Captions: Data Collection and Evaluation Server

  cs.CV · 2015-04-01 · 3 theorem links

  Microsoft COCO Captions provides 1.5 million human captions across 330,000 images and a public server to evaluate captioning models with BLEU, METEOR, ROUGE, and CIDEr.

• Attention U-Net: Learning Where to Look for the Pancreas

  cs.CV · 2018-04-11 · 3 theorem links

  Attention gates added to U-Net automatically focus on target organs in CT images and improve segmentation performance on abdominal datasets.

• ModelScope Text-to-Video Technical Report

  cs.CV · 2023-08-12 · 3 theorem links

  ModelScopeT2V is a 1.7-billion-parameter text-to-video model built on Stable Diffusion that adds temporal modeling and outperforms prior methods on three evaluation metrics.

• DriveVLM: The Convergence of Autonomous Driving and Large Vision-Language Models

  cs.CV · 2024-02-19 · 3 theorem links

  DriveVLM adds vision-language models with scene description, analysis, and hierarchical planning modules to autonomous driving, paired with a hybrid DriveVLM-Dual system tested on nuScenes and SUP-AD datasets and deployed on a production vehicle.

• Motus: A Unified Latent Action World Model

  cs.CV · 2025-12-15 · 3 theorem links

  Motus unifies understanding, video generation, and action in one latent world model via MoT experts and optical-flow latent actions, reporting gains over prior methods in simulation and real robots.

• MMBench: Is Your Multi-modal Model an All-around Player?

  cs.CV · 2023-07-12 · 3 theorem links

  MMBench is a new bilingual benchmark that uses curated questions, CircularEval, and LLM-assisted answer conversion to provide objective, fine-grained evaluation of vision-language models.

• Eliciting Latent Predictions from Transformers with the Tuned Lens

  cs.LG · 2023-03-14 · 3 theorem links

  Training per-layer affine probes on frozen transformers yields more reliable latent predictions than the logit lens and enables detection of malicious inputs from prediction trajectories.

• Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models

  cs.LG · 2025-10-06 · 3 theorem links

  ACE framework evolves contexts for LLMs via modular generation-reflection-curation to outperform baselines by 10.6% on agents and 8.6% on finance while reducing costs and matching top agents with smaller models.

• FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness

  cs.LG · 2022-05-27 · 3 theorem links

  FlashAttention reduces GPU high-bandwidth memory accesses in self-attention via tiling, delivering exact attention with lower IO complexity, 2-3x wall-clock speedups on models like GPT-2, and the ability to train on sequences up to 64K long.

next page →

← previous page