HISTORY近 30 天历史柱高表示当天去重热搜数量
601757185519512056215922592360245625552655276028602962305901570253035704560557065707570853095810541154125513541459155916
09/17—10/16 有历史数据
- 01期刊：Transactions on Machine Learning Research · 摘要：Trajectory-User Linking (TUL) is a critical task in mobility applications that links unlabeled spatial trajectories to the users or entities that generated them. In these applications, data often arrives as a continuous stream and may experience distributional shifts over time. While adapting TUL models via online learning could address these challenges, this approach remains unexplored in current research. Our work bridges this gap by conducting comprehensive evaluations of common TUL techniques in an online learning context. To improve the performance of existing TUL techniques in this sett… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/lucasczz/MiLES · OpenReview ID：YfyOCduclRLucas Cazzonelli, Steffen Thoma
- 02期刊：Transactions on Machine Learning Research · 摘要：This paper studies continual test-time adaptation (CTTA), the task of adapting a model to constantly changing unseen domains in testing while preserving previously learned knowledge. Existing CTTA methods mostly focus on adaptation to the current test domain only, overlooking generalization to arbitrary test domains a model may face in the future. To tackle this limitation, we present a novel online test-time domain generalization framework for CTTA, dubbed TestDG. TestDG aims to learn features invariant to both current and previous test domains on the fly during testing, improving the potent… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/sohyun-l/TestDG · OpenReview ID：qZ1HuY07UdSohyun Lee, Nayeong Kim, Juwon Kang et al.
- 03期刊：Transactions on Machine Learning Research · 摘要：Recent advances in vision-language model (VLM) reasoning have been largely attributed to the rise of reinforcement learning (RL), which has shifted the community’s focus away from the supervised fine-tuning (SFT) paradigm. Many studies suggest that introducing the SFT stage not only fails to improve reasoning ability but may also negatively impact model training. In this study, we revisit this RL-centric belief through a systematic and controlled comparison of SFT and RL on VLM reasoning. Using identical data sources, we find that the relative effectiveness of SFT and RL is conditional and st… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：EWk0VxDpaxYongcan Yu, Lingxiao He, Shuo Lu et al.
- 04期刊：Transactions on Machine Learning Research · 摘要：Conformal regression provides finite-sample marginal coverage, but interval efficiency depends on how uncertainty is allocated across inputs. We propose \emph{Conformal Laplace-Aware Predictive Scaling} (CLAPS), which combines a learned heteroscedastic scale with last-layer Laplace uncertainty. The learned variance also weights the training features used to construct the posterior, yielding a noise-weighted support correction in the learned representation. The correction contracts as weighted feature information increases and becomes most influential under localized support imbalance. Experim… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/AndrewKim1997/CLAPS · OpenReview ID：RvoPfNBjwpDongseok Kim, Hyoungsun Choi, Mohamed Jismy Aashik Rasool et al.
- 05期刊：Transactions on Machine Learning Research · 摘要：In a controlled fixed-budget study of linear attention decay mechanisms, we find that the delta rule is the largest observed factor under our protocol. We evaluate seven linear-attention variants plus a standard softmax baseline on two datasets (TinyStories, WikiText-103) with 3 random seeds each at both 18M and 125M parameters, and 5 variants at 42M on TinyStories. The data-independent subset, where decay gates are learned parameters rather than token-conditioned functions, forms a complete 2×2 comparison over granularity (scalar vs. channel-wise) and delta rule (absent vs. present), while t… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/emailtovamos/delta-rule-linear-attention · OpenReview ID：FdXB4u27LDSatyajit Das
- 06期刊：Transactions on Machine Learning Research · 摘要：Understanding the agent's learning process, particularly the factors that contribute to its success or failure post-training, is crucial for comprehending the rationale behind the agent's decision-making process. Prior methods clarify the learning process by creating a structural causal model (SCM) or visually representing the distribution of value functions. Nevertheless, these approaches have constraints as they exclusively function in 2D environments or with simple transition dynamics. Understanding the agent's learning process in complex environments or tasks is more challenging. In this… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：Dmu7agmmTQShuang Ao, Flora D. Salim, Haris Aziz et al.
- 07期刊：Transactions on Machine Learning Research · 摘要：Deep neural networks (DNNs) and Kolmogorov-Arnold networks (KANs) provide flexible function approximators, but their accuracy can require substantial network depth, width, and trainable parameter counts. Beyond making the resulting networks less transparent, large parameter spaces can increase training cost and produce solutions that are sensitive to optimization and parameter initialization, leading to variability in generalization error. For these reasons, we propose Shallow Universal Polynomial Networks (SUPNs), which retain a single hidden layer while increasing the expressivity of each u… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：srHGkCrSTMZachary Morrow, Michael Penwarden, Brian Chen et al.
- 08期刊：Transactions on Machine Learning Research · 摘要：Known Lie symmetries specify how a partial differential equation (PDE) solution operator should transform under group actions, but finite supervised training data constrain this relationship only at observed input--output pairs. To encourage equivariance without changing the architecture, we introduce \textbf{LOCO} (\textbf{L}ocal \textbf{O}rbit \textbf{Co}nsistency), a lightweight and flexible training objective for neural operators. LOCO samples a small Lie-group transformation during training and explicitly penalizes the mismatch between the operator's prediction on the transformed input a… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/michalmariuszlewandowski/orbit-tanget-no · OpenReview ID：AQIyD6jSSlMichal Lewandowski, Bernhard Heinzl, Roman Rainer et al.
- 09期刊：Transactions on Machine Learning Research · 摘要：Many real-world tasks require optimizing expensive black-box functions accessible only through noisy evaluations, a setting commonly addressed with Bayesian optimization. While Gaussian processes (GPs) and Bayesian neural networks have driven significant progress, GP-based methods scale cubically with observations, and BNN-based approaches remain burdened by expensive posterior sampling such as Hamiltonian Monte Carlo, and inner-loop acquisition optimization in high-dimensional setting. In this work, we propose VBO-MI (Variational Black-Box Optimization with Mutual Information), a fully gradi… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：nkj5zYqJ2pFarhad Mirkarimi
- 10期刊：Transactions on Machine Learning Research · 摘要：In contemporary cloud-based services, protecting users’ sensitive data and ensuring the confidentiality of the server’s model are critical. Fully homomorphic encryption (FHE) enables inference directly on encrypted inputs, but its practicality is hindered by expensive bootstrapping and inefficient approximations of non-linear activations. We introduce Safhire, a hybrid inference framework that executes linear layers under encryption on the server while offloading non-linearities to the client in plaintext. This design eliminates bootstrapping, supports exact activations, and significantly red… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/sacs-epfl/SAFHIRE · OpenReview ID：AAneiBMYgSSayan Biswas, Philippe Chartier, Akash Dhasade et al.
- 11期刊：Transactions on Machine Learning Research · 摘要：Parameter estimation in stochastic differential equations is a classical statistical problem of much importance in many scientific fields. Recent work by Tapia-Costa et al. (2026) introduced a novel technique for estimating the drift when the diffusion parameter is known, using discrete samples from multiple trajectories. Their method treats drift estimation as a denoising problem, and leverages tools from (conditional) score-matching diffusion models. Although their experiments showed promising results across different drift classes, the question of theoretical guarantees for their estimator… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：cyHG8JSTGeIoar Casado Telletxea, Omar Rivasplata
- 12期刊：Transactions on Machine Learning Research · 摘要：Training deep neural networks requires extensive experimentation to estimate how long it will take, and how much it will cost, for a model to reach convergence. Chronos is a unified framework for pre-training prediction of both the convergence horizon and the resulting end-to-end training time and cost for a target configuration. Here, convergence is defined as the epoch at which early stopping is triggered after a specified number of epochs without sufficient improvement in validation loss. Unlike approaches that estimate these quantities from observations collected during an executing train… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：gDLPOEjQTHAlireza Pourali, Arian Boukani, Hamzeh Khazaei
- 13期刊：Transactions on Machine Learning Research · 摘要：Structure-based drug design has been accelerated by pocket-aware 3D generative models, yet most methods primarily fit the training distribution and may fall short of satisfying multiple properties required in real-world therapeutic drug discovery. Recently, increasing attention has focused on structure-based molecule optimization (SBMO), which targets fine-grained control over multiple specified molecular properties. In this paper, we present DEPPA, a novel SBMO approach building upon Denoising Diffusion Policy Optimization for fine-tuning a pre-trained pocket-aware diffusion model via reinfo… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/xy9485/DePPA · OpenReview ID：5sjQt9I2mtYuan Xue, Daniel Kudenko, Megha Khosla
- 14期刊：Transactions on Machine Learning Research · 摘要：Integer programming (IP) is central to many combinatorial optimization tasks but remains challenging due to its NP-hard nature. A practical way to improve IP solvers is to manually design inequality families that exploit problem structure. However, this creative process requires deep expertise and has therefore been difficult to automate. Our proposed framework, Symbolic Acceleration via Generative Evolution (SAGE), automates symbolic discovery of acceleration inequality families at the modeling level. It reasons over a symbolic IP formulation and a natural language description of the problem… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/milad1378yz/sage · OpenReview ID：rxJvINT074Milad Yazdani, Mahdi Mostajabdaveh, Samin Aref et al.
- 15期刊：Transactions on Machine Learning Research · 摘要：Performative prediction (PP) provides a framework for learning in settings where model deployment influences the data distribution on which the model is trained. Unlike conventional learning under a fixed data distribution, PP involves model-induced distribution shifts that make it challenging to design algorithms converging to stable points, known as stationary performatively stable (SPS) solutions. Existing algorithms for finding SPS solutions, including repeated gradient descent (RGD) and greedy stochastic gradient descent (SGD-GD), have largely focused on strongly convex losses. Although… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/osu-srml/SPRINT · OpenReview ID：q0L3a1zSroTian Xie, Ding Zhu, Jia Liu et al.
- 16期刊：Transactions on Machine Learning Research · 摘要：Fine-tuning adapts pretrained models to new tasks, but identifying which weight changes support the resulting behavior remains difficult. Weight-space interpretability assigns semantic descriptions to these changes, called fine-tuning deltas. Such descriptions leave open how much task performance depends on the identified updates, particularly when other updates can support the same predictions. We address this gap by pairing reversion, which restores selected parameters to their pretrained values, with retention, which keeps only selected fine-tuning updates. We evaluate Contrastive Language… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：AQ7a5vGI0vChih-Ting Liao, Kezhen Bao, Xin Cao
- 17期刊：Transactions on Machine Learning Research · 摘要：One-shot federated learning (OSFL) addresses the communication overhead of federated learning by limiting training to a single round, but doing so without sacrificing model quality is non-trivial, particularly when client data distributions diverge. Recent work has addressed this challenge by aggregating client knowledge on the server through the construction of transferable synthetic datasets or distillates. However, most of these methods lack formal privacy guarantees, leaving a gap in jointly achieving low communication, robustness to heterogeneity, and rigorous privacy. We propose FedKT-C… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：9H3b5MX87mMaximilian Andreas Hoefler, Karsten Mueller, Wojciech Samek
- 18期刊：Transactions on Machine Learning Research · 摘要：We propose GENERIC-FNO, a neural operator that places the metriplectic (GENERIC) degeneracy structure of non-equilibrium thermodynamics directly in function space: reversible, energy-conserving dynamics are coupled to irreversible, entropy-producing dynamics via the degeneracy conditions. Structure-preserving neural operators developed previously enforce no more than one conservation law, or else a purely Hamiltonian form, and thermodynamically consistent learning has been restricted to finite-dimensional, graph, or particle systems. In GENERIC-FNO, the energy and entropy functionals are lear… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/jaysulk/GENERIC-FNO · OpenReview ID：gbpEYDdfDBJason Alan Sulskis, Sathya N. Ravi
- 19期刊：Transactions on Machine Learning Research · 摘要：Bayesian neural networks (BNNs) offer a principled framework for quantifying epistemic uncertainty. However, standard weight-space priors yield complex, high-dimensional neural network posteriors that are computationally prohibitive to sample from, and popular approximate inference schemes often fail to capture epistemic uncertainty effectively---a failure mode particularly acute in reinforcement learning (RL) \citep{osband2018randomized}. To resolve this challenge, we introduce a new class of BNNs that quantify uncertainty by placing Bayesian nonparametric Dirichlet Process (DP) priors direc… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/sumilations/Dirichlet-Process-Bayesian-Neural-networks · OpenReview ID：VrWIwB8g3ZSumit Vashishtha
- 20Mechanisms of Multimodal Synchronization: Insights from Decoder-Based Video-Text-to-Speech Synthesis期刊：Transactions on Machine Learning Research · 摘要：Unified decoder-only transformers have shown promise for multimodal generation, yet how sequence design affects the synchronization of modalities with heterogeneous sampling rates remains under explored. We investigate this through video-text-to-speech (VTTS) synthesis---a controlled task requiring fine-grained temporal alignment between sparse text, video, and continuous speech. Using a unified decoder-only transformer, dubbed Visatronic, trained on VoxCeleb2, we study: (i) how modalities contribute complementary information, (ii) how positional encoding strategies enable synchronization acr… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：D5QpMy6yUHAkshita Gupta, Tatiana Likhomanenko, Karren Dai Yang et al.
- 21期刊：Transactions on Machine Learning Research · 摘要：Fine-tuning directly on user-side data is an effective way to improve the performance of widely adopted data-driven human mesh recovery (HMR) models. However, doing so for existing HMR methods often assumes aggregating user-side data from real-world HMR-deployed devices to a central training server, posing a significant privacy risk as sensitive human images are transmitted. How can human mesh recovery be evaluated and improved in privacy-constrained real-world settings? We study this question through a benchmark in which state-of-the-art HMR models are trained with federated learning and a s… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：4XHzBaL0qAZeyu Cao, Qingxuan Wu, Zhiyang Dou et al.
- 22期刊：Transactions on Machine Learning Research · 摘要：Neural ordinary differential equations (NODEs) are geometric deep learning models based on dynamical systems and flows generated by vector fields on manifolds. Despite numerous successful applications, particularly within the flow matching paradigm, all existing NODE models are fundamentally constrained to fixed-dimensional dynamics by the intrinsic nature of the manifold's dimension. In this paper, we extend NODEs to M-polyfolds (spaces that can simultaneously accommodate varying dimensions and a notion of differentiability) and introduce PolyNODEs, the first intrinsic variable-dimensional f… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/turbotage/PolyNODE/tree/tmlr_2026 · OpenReview ID：CKQ1RfnHbnPer Åhag, Alexander Friedrich, Fredrik Ohlsson et al.
- 23期刊：Transactions on Machine Learning Research · 摘要：Chain-of-Thought (CoT) prompting significantly enhances large language models' (LLMs) problem-solving capabilities, but still struggles with complex multi-hop questions, often falling into circular reasoning patterns or deviating from the logical path entirely. This limitation is particularly acute in retrieval-augmented generation (RAG) settings, where obtaining the right context is critical. We introduce Prolog-Initialized Chain-of-Thought ($\pi$-CoT), a novel prompting strategy that combines logic programming's structural rigor with language models' flexibility. $\pi$-CoT reformulates mult… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：gykgXffnE8Chao Wan, Albert Gong, Mihir Mishra et al.
- 24期刊：Transactions on Machine Learning Research · 摘要：Reinforcement Learning with Verifiable Rewards (RLVR) is often framed as a choice between the expressivity of Full Fine-Tuning (FFT) and the efficiency of Low-Rank Adaptation (LoRA). We find that this framing misses an important variable: the geometry of the update. Under a common cold-start GRPO protocol, LoRA exhibits rank collapse: performance remains broadly stable at low rank, then fails abruptly as rank increases. On DeepMath-Hard, for example, accuracy drops from 73.1% at r=64 to 4.7% at r=128. In contrast, Weight-Decomposed Low-Rank Adaptation (DoRA) and Quantum-Informed Tensor Adapta… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/Shrimpstanot/structure-over-scale · OpenReview ID：17kr8cCW7sAllan Kazakov, Abdurrahman Javat
- 25期刊：Transactions on Machine Learning Research · 摘要：We study the \emph{modality gap} in CLIP-style dual-encoder contrastive learning, where image and text embeddings remain misaligned despite being trained in a shared space. \new{We argue that the gap is promoted by the InfoNCE formulation when the two encoders share no parameters.} We conduct a uni-modal experiment with \new{two identically initialized encoders with disjoint parameter sets} and find that InfoNCE \emph{actively} generates a gap at low temperatures. \new{We analyze this phenomenon and show that at low temperatures, and under an irreducible approximation error, InfoNCE can be re… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：Code availability: https://github.com/fmager/modality-gap · OpenReview ID：P0fmivB4EHFabian Mager, Hiba Nassar, Lars Kai Hansen
- 26期刊：Transactions on Machine Learning Research · 摘要：Automated decision-making algorithms drive applications in domains such as recommendation systems and search engines. These algorithms often rely on off-policy contextual bandits or off-policy learning (OPL). Conventionally, OPL selects actions that maximize the expected reward from an existing action set. However, in many real-world scenarios, actions, such as news articles or video content, change continuously, and the action space may differ from that at the time when the logged data were collected. We define actions introduced after deploying the logging policy as new actions and focus on… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/usaito/TMLR2026-new-actions · OpenReview ID：3YTlnocq4PRen Kishimoto, Tatsuhiro Shimizu, Kazuki Kawamura et al.
- 27期刊：Transactions on Machine Learning Research · 摘要：Passive acoustic monitoring (PAM) for bird classification suffers from a severe domain shift between focal training data and passive deployment environments, a challenge exacerbated by the prohibitive cost of annotating target data. Extensive benchmarking has shown that standard architectures and data augmentations struggle to generalize across this focal-to-passive gap. We propose SNR-Guided Multi-Signal Mix (SG-MSM), a multi-stage data augmentation pipeline that explicitly mitigates the focal-to-passive domain shift. We evaluate SG-MSM across six diverse PAM environments using an EfficientN… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/ERR-afk/SG-MSM · OpenReview ID：qUH4rZhABlEZHINI RASENDIRAN R, Chandresh Kumar Maurya
- 28期刊：Transactions on Machine Learning Research · 摘要：Vision-language models (VLMs) fail systematically on counterfactual visual reasoning tasks, defaulting to canonical linguistic priors even when visual evidence directly contradicts them. Prior work documents this phenomenon but does not explain where the failure originates—in visual perception or in language-level reasoning. We address this gap through a controlled two-condition ablation study across 13 VLMs (spanning 2.2--235B parameters, 2023--2025) on a matched dataset of 60 counterfactual images and 60 structurally equivalent scene graphs. We establish three findings. First, semantic bias… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/Imsachin010/do_vlms_actually_see · OpenReview ID：o9KC3v0pvbSachin Mishra, Meenakshi D'Souza
- 29期刊：Transactions on Machine Learning Research · 摘要：Principal component analysis (PCA) is one of the most well-studied machine learning methods of the last century. However, the principal components derived by PCA are not guaranteed to be response-informative and are usually dense, meaning they are hard to interpret in high dimensional settings. The former has led to the development of supervised PCA techniques where the response is usually incorporated in an objective function to guide informative projections and enhance predictive accuracy, while the latter to sparse PCA methods that seek to induce sparsity by shrinking non-significant varia… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/theopapazoglou/SCSPCA · OpenReview ID：c2XDHSfZt2Theodosios Papazoglou, Guosheng Yin
- 30期刊：Transactions on Machine Learning Research · 摘要：Causal bandits exploit structural relationships among variables to share information across interventions and accelerate the identification of high-reward decisions. In many applications, however, some variables cannot be directly manipulated, even though they influence the reward and provide useful information about the underlying causal system. We study contextual causal bandits with non-manipulable variables, where context variables are observed before action selection and additional variables are observed after each intervention. Assuming a known causal graph without latent confounding, w… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/QasimElahi/Code_for_Paper_Information-Directed-Sampling-for-Causal-Bandits_TMLR2026 · OpenReview ID：HCJMKOxGjWMuhammad Qasim Elahi, Murat Kocaoglu, Mahsa Ghasemi
- 31期刊：Transactions on Machine Learning Research · 摘要：Large language models (LLMs) often map semantically related prompts to concentrated internal representations at particular layers, even when their surface forms differ quite a bit. We study this finite-depth phenomenon through an Iterated Function System (IFS)-inspired model of concept-conditioned representation dynamics. Rather than assuming that individual transformer blocks are contractive, we give local conditions under which a finite range of layers will contract concept-conditioned trajectories. Our construction suggests where to read a concept (i.e., depth at which it is most concentra… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：N9bm8nADFDSotirios Panagiotis Chytas, Vikas Singh
- 32期刊：Transactions on Machine Learning Research · 摘要：Flat minima are strongly associated with improved generalisation in deep neural networks. However, this connection has proven nuanced in recent studies, with both theoretical counterexamples and empirical exceptions emerging in the literature. In this paper, we revisit the role of sharpness in model performance and examine sharpness as a function-dependent property rather than as a direct indicator of poor generalisation. We conduct extensive empirical studies ranging from single-objective optimisation, synthetic non-linear binary classification tasks, to modern image classification tasks. In… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/IFMW01/a_function-centric_perspective_on_flat_and_sharp_minima · OpenReview ID：LV1ffpXXeVIsrael Mason-Williams, Gabryel Mason-Williams, Helen Yannakoudakis
- 33期刊：Transactions on Machine Learning Research · 摘要：We propose Squeeze3D, a novel framework that leverages implicit prior knowledge learnt by existing pre-trained encoders and decoders to compress 3D data at extremely high compression ratios. Our approach bridges the latent spaces between a pre-trained encoder and a pretrained decoder model through trainable mapping networks. Any 3D asset represented as a mesh, point cloud, or radiance field is first encoded by the pre-trained encoder and then transformed (i.e. compressed) into a highly compact latent code by a mapping network. This latent code can effectively be used as an extremely compresse… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://squeeze3d.github.io/ · OpenReview ID：XXYGHfqzvARishit Dagli, Yushi Guan, Sankeerth Durvasula et al.
- 34期刊：Transactions on Machine Learning Research · 摘要：Modern machine learning increasingly spans supervised, transfer, continual, meta-learning, and other training regimes, often while reusing the same hypothesis families, architectures, and optimizers. Yet these regimes can differ substantially in information access, success criteria, memory assumptions, adaptation protocols, and sample accounting. As a result, it is often unclear whether a proposed paradigm is genuinely distinct, a special case of an existing paradigm, or part of a broader structural relationship among paradigms. This motivates a structural theory for comparing learning paradi… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：8ZyXUFRfb2Junwei Su, Changjie Wang, Dongyang Chang
- 35期刊：Transactions on Machine Learning Research · 摘要：Computer-Aided Design (CAD) models are defined by construction histories: parametric recipes that encode design intent. Yet large-scale 3D datasets mostly provide only boundary representations or meshes, omitting this procedural information. We introduce Zero-to-CAD, a scalable framework that casts CAD synthesis as agentic search: a large language model writes, executes, validates, and repairs code within a tool-equipped CAD environment. At scale, the framework produces approximately one million executable, readable, and editable construction sequences with a broad operation vocabulary beyond… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://huggingface.co/collections/ADSKAILab/zero-to-cad · OpenReview ID：QiKZ2TPGL0Mohammadmehdi Ataei, Farzaneh Askari, Kamal Rahimi Malekshan et al.
- 36期刊：Transactions on Machine Learning Research · 摘要：Large language models struggle with factual hallucinations, and mitigating them typically requires executing the model to assess uncertainty. We introduce a query-dependent rate--distortion framework showing that factual accuracy follows a predictable, sigmoidal ``knowledge cliff'' governed by exposure-linked entity prominence. Below a critical frequency $f_{\mathrm{crit}}$, the model rarely recalls facts reliably; above this threshold, accuracy rises sharply. We present five core findings based on this framework. First, mapping this frequency response yields an accurate, pre-inference risk s… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：QG0g950T1tNevroz Sen
- 37期刊：Transactions on Machine Learning Research · 摘要：Many methods for distribution shift under causal structure assume the sparse mechanism shift (SMS) hypothesis: that across environments only a few causal conditionals change. This assumption drives mechanism-shift scoring, causal discovery in heterogeneous data, and transportable prediction, yet it is almost never tested on the data at hand. This paper makes SMS diagnosable. We first ask why it is hard to tell which mechanisms changed once the causal graph must be estimated rather than assumed known. A controlled ablation locates the cause: the false positives that limit precision arise at th… · 篇幅：Beyond PDF submission (pageless, webpage-style content) · 代码：https://github.com/Song1823/SparseMechanismShift · OpenReview ID：6AGOgvdze1Song Shuang, Muhammad Syafiq Mohd Pozi, Nik Fatinah N. Mohd Farid
- 38期刊：Transactions on Machine Learning Research · 摘要：Reinforcement learning (RL) has improved the reasoning abilities of large language models (LLMs), yet state-of-the-art methods still fail to learn on many training problems. On hard problems, on-policy RL rarely explores even a single correct rollout, yielding zero reward and no learning signal. We find that natural solutions to remedy this exploration problem from classical RL, such as entropy bonuses, more permissive clipping of the importance ratio, or direct optimization of pass@k objectives, do not resolve this issue and often destabilize optimization without improving solvability. A nat… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：pbyIoUW7YVYuxiao Qu, Amrith Setlur, Virginia Smith et al.
- 39期刊：Transactions on Machine Learning Research · 摘要：As modern machine-learning models are deployed in high-stakes, data-rich environments, interactions among features have become more intricate and less amenable to traditional interpretation. Strong feature dependence can make attribution sensitive to how shared predictive information is allocated among correlated predictors. In the presence of mul- ticollinearity or near-duplicate predictors, attribution tools such as SHAP, LIME, HSIC, MI/CMI, and SAGE may distribute importance across redundant features, obscuring which variables contain unique predictive information. This can produce unstabl… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/Poushali96/MCIR · OpenReview ID：nRQhWIzaM5Poushali Sengupta, Sabita Maharjan, Frank Eliassen et al.
- 40期刊：Transactions on Machine Learning Research · 摘要：Classifier-Free Guidance (CFG) with guidance scale $\gamma$ is commonly motivated by the $\omega$-powered conditional distribution obtained by setting $\omega=\gamma$, but CFG does not, in general, sample from this distribution exactly. We study whether the powered target can instead be learned explicitly. We introduce Importance-Weighted Conditional Flow Matching (IW-CFM), which reweights conditional flow-matching examples using density ratios obtained from frozen conditional and unconditional flow models. With exact normalized weights, the resulting objective has the same gradient as flow m… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://bitbucket.org/dmmlgeneva/iw-cfm · OpenReview ID：fBli1qL3ddMaciej Falkiewicz, Frantzeska Lavda, Jason Ramapuram et al.
- 41期刊：Transactions on Machine Learning Research · 摘要：The Segment Anything Model (SAM) family has become a widely adopted vision foundation model, but its ability to control segmentation granularity remains limited. Users often need to refine results manually — by adding more prompts or selecting from pre-generated masks — to achieve the desired level of detail. This process can be ambiguous, as the same prompt may correspond to several plausible masks, and collecting dense annotations across all granularities is prohibitively expensive, making supervised solutions infeasible. To address this limitation, we introduce UnSAMv2 which enables segmen… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/yujunwei04/UnSAMv2 · OpenReview ID：TJyPRRrLIVJunwei Yu, Trevor Darrell, XuDong Wang
- 42期刊：Transactions on Machine Learning Research · 摘要：Training large language models (LLMs) for multi-turn code generation typically relies on online reinforcement learning (RL), in which models learn from repeatedly generated online rollouts. Although effective, online RL can be costly and unstable. In this paper, we present contextual bandit learning with offline trajectories (COBALT), a simple method that reuses offline multi-turn trajectories to efficiently train LLMs for multi-turn code generation. Building on the observation that multi-turn code generation is a one-step recoverable Markov decision process, COBALT first collects code genera… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：7SwoUNzIxeZiru Chen, Dongdong Chen, Ruinan Jin et al.
- 43期刊：Transactions on Machine Learning Research · 摘要：Large Language Models struggle to reliably acquire new knowledge post-deployment---even when relevant text resources exist, models fail to transform them into actionable knowledge without retraining. Retrieval-Augmented Generation attempts to bridge this gap by surfacing relevant documents at inference time, yet similarity-based retrieval often fails to identify context that actually improves task performance. We introduce Evolutionary Context Search (ECS), an evolutionary method that searches over context combinations guided accuracy on a small development set, requiring only inference calls… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：nedfqQFmbHQi Sun, Stefan Nielsen, Rio Yokota et al.
- 44期刊：Transactions on Machine Learning Research · 摘要：3D Gaussian Splatting represents scenes as finite mixtures of anisotropic Gaussians whose number of components $K$ is governed by heuristics or user-specified caps. Variational Bayes Gaussian Splatting (VBGS) casts splat fitting as conjugate variational inference with a finite Dirichlet weight prior and fixed $K$. DP-Splat replaces that prior with either a truncated stick-breaking Dirichlet-process prior or a sparse overfitted Dirichlet, so the number of occupied components can adapt to the data while every coordinate-ascent update stays in closed form. As in VBGS, the likelihood is a mixture… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/archiedong/dp-splat · OpenReview ID：75Hx0RDPMrAqi Dong
- 45期刊：Transactions on Machine Learning Research · 摘要：In this work, we introduce SOARA (Subspace Orthogonal Adaptation via Rotational Alignment), a novel family of parameter-efficient fine-tuning (PEFT) algorithms that navigate the manifold of pretrained weights through geometric alignment. While existing SVD-based methods prioritize magnitude-driven adaptation, for example by modifying singular values (SALT, SVFT) or reparameterizing basis vectors via additive low-rank updates (PiSSA), they often overlook the intrinsic rotational symmetries of the feature space. In contrast, SOARA treats the pretrained principal singular vectors as a fixed basi… · 篇幅：Long submission (more than 12 pages of main content) · 代码：https://github.com/chandan22140/soara-peft · OpenReview ID：gvh5FQXr9wChandan Sah, Vinayak Abrol, Anubha Gupta
- 46期刊：Transactions on Machine Learning Research · 摘要：In real-world applications of multivariate time series classification (TSC), distribution shift between training and testing data is common, often leading to degraded out-of-distribution (OOD) performance relative to in-distribution (ID) performance. Existing methods typically improve robustness through the training objective or data augmentation, and evaluate models that use early fusion, where channels are jointly processed. However, the impact of fusing channels at later stages (middle and late fusion) has not been studied in isolation. We show that later fusion structurally isolates chann… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：lbAzMRlrFCTom Ryder, Xi Chen
- 47期刊：Transactions on Machine Learning Research · 摘要：Modern label aggregation increasingly feeds soft posteriors into downstream learners such as preference and reward models, making posterior calibration as important as top-1 accuracy. In sparse crowdsourcing, however, each item receives only a few annotations from workers who may share systematic biases, so correlated agreement can be mistaken for independent evidence and transformed into overconfident posteriors. We identify this failure mode as illusory evidence accumulation and propose BRAVE, a block-wise structural regularization framework for reliable label aggregation. BRAVE represents… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/zerunniu/BRAVE · OpenReview ID：iWFI5hO1dZZerun Niu, Junming Huang, Long Tan Le et al.
- 48期刊：Transactions on Machine Learning Research · 摘要：Large Language Models (LLMs) are increasingly applied in high-stakes domains such as medicine and law, where hallucinations can have serious consequences. Existing detection approaches either depend on costly proprietary LLMs with limited adaptability, or on monolithic open-source models that require full retraining, struggle with long evidence contexts, and lack transparency. We introduce LiteHall, a lightweight, fully open-source, three-stage hallucination detection pipeline designed for modularity, domain adaptability, and interpretability. Each stage leverages a 1.7B-parameter Small Langu… · 篇幅：Regular submission (no more than 12 pages of main content) · OpenReview ID：TRR4EUnKmDJaladi Sri Harsha, Komuravelli Prashanth, Arun Kumar S
- 49期刊：Transactions on Machine Learning Research · 摘要：Given a dataset of finitely many elements $\mathcal{T} = \{\mathbf{x}_i\}_{i = 1}^N$, the goal of dataset condensation (DC) is to construct a synthetic dataset $\mathcal{S} = \{\tilde{\mathbf{x}}_j\}_{j = 1}^M$ which is significantly smaller ($M \ll N$) such that a model trained from scratch on $\mathcal{S}$ achieves comparable or even superior generalization performance to a model trained on $\mathcal{T}$. Recent advances in DC reveal a close connection to the problem of approximating the data distribution represented by $\mathcal{T}$ with a reduced set of points. In this work, we present a… · 篇幅：Long submission (more than 12 pages of main content) · OpenReview ID：xByeYLCMMGTong Chen, Raghavendra Selvan
- 50期刊：Transactions on Machine Learning Research · 摘要：The amount of quality data in many machine learning tasks is limited to what is available locally to data owners. The set of quality data can be expanded through trading or sharing with external data agents. However, external data may be contaminated or introduce undesirable sample diversity which can degrade performance of personalized machine learning tasks, as in diagnosis of a rare disease or recommendation systems. Therefore, data buyers need quality guarantees prior to data acquisition. Previous works primarily rely on distributional assumptions about data from different agents, relegat… · 篇幅：Regular submission (no more than 12 pages of main content) · 代码：https://github.com/Martin497/ConformalDataContaminationTest · OpenReview ID：7HNbXnhtfRMartin V. Vejling, Shashi Raj Pandey, Christophe A. N. Biscio et al.