A Note on Code Quality Score: LLMs for Maintainable Large Codebases
Comments 24 pages, ICLR format
期刊&会议
ICLR
International Conference on Learning Representations · 会议 · Machine Learning
Boost Self-Supervised Dataset Distillation via Parameterization, Predefined Augmentation, and Approximation
机构 * National Yang Ming Chiao Tung University（国家阳明交通大学） ； Macronix International Co., Ltd.（宏碁国际公司）
Comments To appear in the Proceedings of the International Conference on Learning Representations (ICLR 2025)
Long-tailed Adversarial Training with Self-Distillation
机构 * Korea Advanced Institute of Science and Technology (KAIST)（韩国科学技术院）
Comments ICLR 2025. See OpenReview and code (in Supplementary Material) at https://openreview.net/forum?id=vM94dZiqx4
Transformer Meets Twicing: Harnessing Unattended Residual Information
机构 * Department of Mathematics National University of Singapore（数学系国立新加坡大学）
Comments 10 pages in the main text. Published at ICLR 2025
Effects of Feature Correlations on Associative Memory Capacity
机构 * Department of Computer Science and Electrical Engineering University of California, Berkeley（计算机科学与电气工程系加州大学伯克利分校）
Comments Accepted at ICLR 2025 "New Frontiers in Associative Memories" Workshop. Code: https://github.com/stefanbielmeier/feature-correlations-am
Range, not Independence, Drives Modularity in Biologically Inspired Representations
机构 * Department of XXX, University of YYY, Location, Country（XXX系，YYY大学，地点，国家） ； School of ZZZ, Institute of WWW, Location, Country（ZZZ学院，WWW研究所，地点，国家）
Comments 37 pages, 12 figures. WD and KH contributed equally; LH and JHL contributed equally
Journal ref Proceedings of the 13th International Conference on Learning Representations, 2025
InstructLayout: Instruction-Driven 2D and 3D Layout Synthesis with Semantic Graph Prior
机构 * Wangxuan Institute of Computer Technology, Peking University（北京大学王轩计算机技术研究所） ； PICO AI group, ByteDance（字节跳动PICO AI团队）
Comments Accepted to T-PAMI 2025. This paper is an extension of ICLR 2024 "InstructScene: Instruction-Driven 3D Indoor Scene Synthesis with Semantic Graph Prior". arXiv admin note: substantial text overlap with arXiv:2402.04717
Weakly Supervised Virus Capsid Detection with Image-Level Annotations in Electron Microscopy Images
机构 * Ulm University（乌尔姆大学） ； TU Vienna（维也纳技术大学）
Journal ref The Twelfth International Conference on Learning Representations. 2024
Improved Algorithms for Kernel Matrix-Vector Multiplication Under Sparsity Assumptions
机构 * MIT（麻省理工学院） ； EPFL（瑞士联邦理工学院） ； Tel Aviv University（特拉维夫大学）
Comments Published in ICLR 2025
X-NeMo: Expressive Neural Motion Reenactment via Disentangled Latent Attention
机构 * Tsinghua University（清华大学） ； ByteDance Inc.（字节跳动公司）
Comments ICLR 2025, code is available at https://github.com/bytedance/x-nemo-inference
Hyperbolic Genome Embeddings
机构 * Columbia University（哥伦比亚大学）
Comments 30 pages, 16 figures, 10 tables. Camera-ready version for ICLR 2025
Journal ref In Proceedings of the Thirteenth International Conference on Learning Representations (ICLR 2025). https://openreview.net/forum?id=NkGDNM8LB0
Exploring Adaptive Structure Learning for Heterophilic Graphs
机构 * Department of Mathematical Sciences（数学科学系） ； Indian Institute of Technology (BHU)（印度理工学院（BHU））
Comments Initially submitted this draft at Tiny ICLR 2025
Geometry of Neural Reinforcement Learning in Continuous State and Action Spaces
机构 * Brown University（布朗大学） ； Amazon Web Services（亚马逊网络服务） ； Department of Computer Science（计算机科学系）
Comments Proceedings of the Thirteenth International Conference on Learning Representations (ICLR 2025). arXiv admin note: text overlap with arXiv:2301.00009
Approximating Full Conformal Prediction for Neural Network Regression with Gauss-Newton Influence
Comments Accepted at the 13th International Conference on Learning Representations (ICLR 2025)
Learning to Clarify: Multi-turn Conversations with Action-Based Contrastive Self-Training
机构 * Google（谷歌） ； Columbia University（哥伦比亚大学）
Comments ICLR 2025; Code: https://github.com/google-research/google-research/tree/master/learning_to_clarify
Overcoming Slow Decision Frequencies in Continuous Control: Model-Based Sequence Reinforcement Learning for Model-Free Control
机构 * Manning College of Information and Computer Sciences University of Massachusetts Amherst（信息与计算机科学学院 马萨诸塞大学阿默斯特分校）
Comments 30 pages, 14 figures, 7 tables. Presented at the Thirteenth International Conference on Learning Representations (ICLR 2025), Singapore, April 24-28, 2025
Improving Reasoning Performance in Large Language Models via Representation Engineering
机构 * Department of Computer Science, IT University of Copenhagen（计算机科学系，丹麦技术大学）
Comments Has been accepted at "The Thirteenth International Conference on Learning Representations (ICLR 2025)" Link to publication: https://openreview.net/forum?id=IssPhpUsKt
Journal ref 13th International Conference on Learning Representations (ICLR 2025)
Fine-Tuned Language Models Generate Stable Inorganic Materials as Text
机构 * NYU（纽约大学） ； Meta FAIR
Comments ICLR 2024. Code available at: https://github.com/facebookresearch/crystal-llm
Boosting Ray Search Procedure of Hard-label Attacks with Transfer-based Priors
Comments Published at ICLR 2025 (Spotlight paper)
ACMP: Allen-Cahn Message Passing with Attractive and Repulsive Forces for Graph Neural Networks
机构 * Shanghai Jiao Tong University（上海交通大学） ； University of New South Wales（新南威尔士大学） ； King Abdullah University of Science and Technology（卡塔尔国王 Abdullah 科学与技术大学） ； Shanghai Jiao Tong University Shanghai Artificial Intelligence Laboratory（上海交通大学上海人工智能实验室）
Comments 26 pages, 5 figures. NeurIPS 2022 Workshop on GLFrontiers (Oral). ICLR 2023 (Spotlight)
Watermark Anything with Localized Messages
机构 * Meta FAIR ； CMAP, École polytechnique（CMAP，巴黎高等理工学院） ； Centre Inria de l’Université de Rennes（里昂大学Inria研究中心）
Comments ICLR 2025
Graph Neural Networks Gone Hogwild
机构 * Department of Computer Science（计算机科学系） ； Princeton University（普林斯顿大学）
Journal ref Proceedings of the International Conference on Learning Representations (ICLR), 2025
Visually Guided Decoding: Gradient-Free Hard Prompt Inversion with Language Models
机构 * Seoul National University（首尔国立大学） ； Sungkyunkwan University（庆熙大学）
Comments ICLR 2025 (Official Code: https://github.com/DonghoonKim-1938/VGD)
Measuring Leakage in Concept-Based Methods: An Information Theoretic Approach
机构 * ETH Zurich Medical Data Science group（苏黎世联邦理工学院医疗数据科学组） ； ETH Zurich（苏黎世联邦理工学院）
Comments Published at ICLR 2025 Workshop on XAI4Science
Benchmarking Mobile Device Control Agents across Diverse Configurations
Comments Accepted to ICLR 2024 Workshop on Generative Models for Decision Making (Spotlight) and CoLLAs 2025
Generalized Consistency Trajectory Models for Image Manipulation
机构 * KAIST（韩国科学技术院）
Comments ICLR 2025 (poster)
Attention with Markov: A Framework for Principled Analysis of Transformers via Markov Chains
机构 * EPFL（苏黎世联邦理工学院） ； UT Austin（得克萨斯大学奥斯汀分校）
Comments Published at ICLR 2025 under the title "Attention with Markov: A Curious Case of Single-Layer Transformers"
A Denoising VAE for Intracardiac Time Series in Ischemic Cardiomyopathy
机构 * Dept. of Computer Science ETH Zurich（计算机科学系，苏黎世联邦理工学院） ； Dept. of Medicine Stanford University（医学系，斯坦福大学） ； Dept. of Cardiology Amsterdam University（心脏病学系，阿姆斯特丹大学）
Comments 9 pages, 2 figures, 3 tables, the last two authors are shared senior authors
Journal ref International Conference on Learning Representations (ICLR) 2024 - Time Series for Healthcare Workshop
Divide and Translate: Compositional First-Order Logic Translation and Verification for Complex Logical Reasoning
机构 * KAIST（韩国科学技术院） ； MIT（麻省理工学院）
Comments ICLR 2025 camera-ready version
Journal ref The Thirteenth International Conference on Learning Representations (ICLR 2025)
Once-for-All: Controllable Generative Image Compression with Dynamic Granularity Adaptation
Comments Accepted by ICLR 2025. Code is available at https://github.com/lianqi1008/Control-GIC
