Machine Learning
Position: Zeroth-Order Optimization in Deep Learning Is Underexplored, Not Underpowered
Sijia Liu, Yicheng Lang, Soumyadeep Pal et al. (+6 more)
Zeroth-order (ZO) optimization, learning from finite differences of function evaluations without backpropagation, has recently regained
attention in deep learning due to its
memory efficiency and applicability to gray- or black-box pipelines. Yet, ZO methods are often dismissed as fundamentally unscalable because of
estimator variance and unfavorable query complexity. We argue that this conclusion might be misguided: ZO optimization is underexplored, not underpowered. We show that many perceived limitations stem from myopic development practices, most notably full-space, element-wise,
estimator-centric designs. We articulate six positions spanning the algorithmic, systems, and evaluation stack. First, we revisit the feasibility boundaries of
estimator-centric ZO methods through variance control, variance-query tradeoffs, and directional-derivative lenses. Then, we identify three underexplored opportunities: (i) subspace and spectral views of ZO that enable interpretable variance reduction with graceful query scaling, (ii) the forward-only nature of ZO as a systems advantage for communication-efficient, pipeline-friendly, and resource-constrained training, and (iii) the need to de-obfuscate ZO evaluations from task complexity. We strongly advocate rethinking ZO optimization around its unique strengths and acting accordingly, opening a viable path toward large-scale, system-aware, and resource-efficient learning with ZO optimization.
Read the full abstract ▾
All authors Sijia Liu, Yicheng Lang, Soumyadeep Pal, Changsheng Wang, Yancheng Huang, Chongyu Fan, James Diffenderfer, Bhavya Kailkhura, Yihua Zhang
Machine Learning · Cross-listed from stat.ML
Pessimistic Risk-Aware Policy Learning in Contextual Bandits
Yilong Wan, Yuqiang Li and Xianyi Wu
We study risk-aware offline policy learning, aiming to learn a decision rule from logged data that is optimal under general risk criteria. This problem is crucial in high-stakes domains where online interaction is infeasible and adverse outcomes must be carefully controlled. However, existing literature on offline contextual bandits either centers on expected-reward criteria or restricts risk considerations to policy evaluation instead of optimization. In this work, we propose a unified distributional framework for optimizing Lipschitz-continuous risk functionals, a broad class of risk measures encompassing mean-variance, entropic risk, and conditional value-at-risk, among others. By developing novel empirical concentration inequalities for importance sampling-based distributional
estimators, our analysis derives data-dependent suboptimality bounds with an $\tilde{\mathcal{O}}(1/\sqrt{n})$ rate, without relying on restrictive uniform overlap assumptions. This rate is minimax optimal and matches that of risk-neutral offline policy optimization, indicating that optimizing general Lipschitz risk criteria incurs no additional statistical cost relative to the expected-reward.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.CV
Neutral-Reference Prompting for Vision-Language Models
Senmao Tian, Xiang Wei and Shunli Zhang
Efficient transfer learning of
vision-
language models (VLMs) commonly suffers from a Base-New Trade-off (BNT): improving performance on unseen (new) classes often degrades accuracy on known (base) classes. Addressing how to boost recognition of unseen classes without sacrificing known-class performance remains a central challenge. Existing work often simplistically attributes the BNT to overfitting on known classes. We observe an interesting phenomenon: VLMs frequently exhibit asymmetric confusion on certain downstream data, i.e., samples of class A are systematically mispredicted as class B, while the reverse confusion (B to A) rarely occurs. For known classes, this kind of bias can be mitigated by tuning using a cross-
entropy loss, but for unseen classes, such pretraining-induced bias persists and harms generalization. Motivated by this, we propose NeRP, a plug-and-play prompting correction strategy that improves discrimination on unseen classes without modifying model parameters. NeRP leverages neutral text
prompts and reference images to measure class-wise prior preferences along the pre-trained inter-class geometry, and combines them with the sample likelihood to obtain the model's surrogate score. If, for a given sample, the prior strongly favors the current prediction while the observed evidence is clearly insufficient, we perform a local flip between easily confusable class pairs, thereby correcting prior-dominated mispredictions. Extensive experiments across multiple backbones and 15 few-shot and cross-domain
benchmarks show that NeRP substantially improves accuracy on unseen classes while preserving known-class prediction performance.
Read the full abstract ▾
Machine Learning
Transformer-like Inference from Optimal Control
Aditya Kudre, Heng-Sheng Chang and Prashant G. Mehta
Decoder-only
transformers compute the conditional probability of the next
token from a sequence of past observations. This paper derives, from first principles, inference architectures that solve the same prediction problem - and in doing so, recovers
transformer-like layer operations as a consequence of optimal control theory. The framework is developed for two model classes: a nonlinear model of discrete-valued processes, directly motivated by the
transformer, and a linear Gaussian model as a tractable baseline. For both model classes, the prediction objective is reformulated as an optimal control problem whose solution yields an explicit inference
algorithm, the dual filter, with a layer structure that mirrors the layer structure of a decoder-only
transformer. Numerical experiments provide a comparison of the optimal control to
attention weights from a trained
transformer. These experiments reveal that when the
embedding dimension is insufficient, the
transformer implicitly exploits non-Markovian structure.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.CL
Syntax Without Semantics: Teaching Large Language Models to Code in an Unseen Language
Vinayshekhar Bannihatti Kumar, Disha Makhija, Manoj Ghuhan Arivazhagan et al. (+1 more)
Large
language models (LLMs) achieve high pass rates on code generation
benchmarks, yet whether they can transfer this ability to
languages absent from pretraining remains poorly understood. We introduce PyLang, a minimal imperative
language absent from all pretraining corpora, and evaluate frontier models zero-shot and fine-tuned Qwen3 (4B, 8B, 32B) on 352 problems. We find that fine-tuning quickly teaches syntax but fails to transfer semantic competence: Python outperforms PyLang by up to 19% across all configurations, and no intervention (multi-task learning, preference tuning, code infilling, or latent-space objectives) closes the gap. An LLM judge reveals that frontier models select an identical
algorithm to Python 80% of the time, yet cannot translate it into a working PyLang implementation., and CKA analysis confirms that fine-tuned models converge to nearly identical internal representations across
languages (CKA > 0.97) while diverging at the output stage. We term this the implementation fidelity gap: models possess
language-agnostic algorithmic understanding but cannot express it in an unfamiliar
language. Our findings highlight the need for training methods that decouple
reasoning from
language-specific realization.
Read the full abstract ▾
All authors Vinayshekhar Bannihatti Kumar, Disha Makhija, Manoj Ghuhan Arivazhagan, Rashmi Gangadharaiah
Machine Learning
VSPO: Vector-Steered Policy Optimization for Behavioral Control
Xuechen Zhang, Zijian Huang, Kai Yang et al. (+3 more)
Modern
language models often need to optimize a primary accuracy objective while also accommodating secondary behavioral preferences, such as verbosity, agreeableness, or the level of technical expertise in its response. In practice, a base model may exhibit a desired behavior very rarely or not at all. Thus, endowing the model with a target behavior creates a sparse behavioral reward bottleneck. To address such multi-objective problems, we introduce Vector-Steered Policy Optimization (VSPO) which employs a steering vector associated with the target behavior to control the behavior intensity of the generated rollouts. VSPO is obtained by modifying GRPO to sample rollouts with varying steering intensities. This process can be interpreted as an on-policy latent self-
distillation procedure where the model internalizes its steering vector. By varying steering intensities, VSPO upsamples rare behaviors and enriches rollout diversity, which alleviates the sparse reward issue and provably accelerates the policy optimization. Through comprehensive theory and experiments, we establish that VSPO has favorable properties compared to vanilla reward shaping and other alternative approaches. Specifically, under a bandit abstraction, VSPO provably achieves better iteration complexity than reward-shaped GRPO when the steering-induced distributions are sufficiently aligned with the target behavior. We evaluate VSPO across multiple
reasoning benchmarks, including MATH and MMLU-Pro, for four target behaviors: explanation expertise, confidence expression, robustness to misleading context, and response verbosity. Our results show that VSPO consistently improves the control along target behavior while maintaining or improving task accuracy compared with reward shaping, teacher-trace
distillation, and guidance-based baselines.
Read the full abstract ▾
All authors Xuechen Zhang, Zijian Huang, Kai Yang, Weijia Zhang, Jiasi Chen, Samet Oymak
Machine Learning
Offline Reinforcement Learning with Universal Horizon Models
Hojun Chung, Junseo Lee and Songhwai Oh
Model-based reinforcement learning (RL) offers a compelling approach to offline RL by enabling value learning on imagined on-policy trajectories. However, it often suffers from compounding errors due to repeated model inference on self-generated states. While
geometric horizon models (GHM) alleviate this issue through direct prediction over a discounted infinite-horizon future, they remain challenged in accurately modeling distant future states. To this end, we introduce universal horizon models (UHM), a generalization of GHM that directly predicts future states under arbitrary horizons. Leveraging this flexibility, we propose a scalable value learning method that employs a winsorized horizon distribution to stabilize training by capping excessively large horizons. Experimental results on 100 challenging OGBench tasks demonstrate that the proposed method outperforms competitive baselines, particularly on tasks with highly suboptimal
datasets and those requiring long-horizon
reasoning. Project page: https://rllab-snu.github.io/projects/UHM/
Read the full abstract ▾
Machine Learning · Cross-listed from cs.CV
CM-EVS: Sparse Panoramic RGB-D-Pose Data for Complete Scene Coverage
Jiale Liu, Jungang Li, Jieming Yu et al. (+12 more)
Modern 3D visual learning relies on observations sampled from metric 3D assets, yet existing scans, meshes, point clouds, simulations, and reconstructions do not directly provide a sparse, comparable, and geometry-consistent panoramic training interface. Dense trajectories duplicate nearby views, source-specific rendering policies yield heterogeneous annotations, and sparse heuristics may miss important regions or introduce depth-inconsistent observations. We study how to convert 3D assets into sparse panoramic RGB-D-pose data that preserves complete scene coverage with low redundancy and auditable provenance. We propose COVER (Coverage-Oriented Viewpoint curation with ERP Range-depth warping), a training-free ERP viewpoint curator that projects geometry observed from selected views into candidate ERP probes, scores incremental coverage, and penalizes depth conflicts. Under bounded proxy error, its greedy coverage proxy preserves the standard coverage-style approximation behavior up to an additive error term. Using COVER, we build CM-EVS (Coverage-curated Metric ERP View Set), a panoramic RGB-D-pose
dataset with 36,373 curated ERP frames from 1,275 indoor scenes across Blender indoor, HM3D, and ScanNet++, complemented by outdoor panoramas from TartanGround and OB3D re-encoded into the same schema. Each frame provides full-sphere RGB, metric range depth, calibrated pose; COVER-produced indoor frames include per-step provenance logs. With a median of only 25 frames per indoor scene, CM-EVS covers all 13 unified room types while maintaining compact scene-level coverage. Experiments show that COVER improves the coverage-conflict trade-off, making CM-EVS a sparse, compact, and auditable RGB-D-pose resource for geometry-consistent panoramic 3D learning.
Read the full abstract ▾
All authors Jiale Liu, Jungang Li, Jieming Yu, Xinglin Yu, Zihao Dongfang, Keyu Shi, Zongjian Ding, Jiahuan Zhang, Shunwen Bai, Haoran Huang, Yurun Wang, Yanxi Wu, Ningzhe Yu, Yudong Gao, Mingjun Cheng
Machine Learning · Cross-listed from cs.CL
Calibrating LLMs with Semantic-level Reward
Fengfei Yu, Ruijia Niu, Dongxia Wu et al. (+2 more)
As large
language models (LLMs) are deployed in consequential settings such as medical question answering and legal
reasoning, the ability to estimate when their outputs are likely to be correct is essential for safe and reliable use, requiring well-calibrated uncertainty. Standard reinforcement learning with verifiable rewards (RLVR) trains models with a binary correctness reward that is indifferent to confidence, providing no penalty for confident but wrong predictions and thereby degrading calibration. Recent work addresses this by training models to produce verbalized confidence scores alongside answers and rewarding agreement with correctness. However, verbalized confidence is calibrated at the
token level and thus exhibits inconsistency across textual variations with same semantic meaning. We propose Calibration with Semantic Reward (CSR), a framework that calibrates
language models directly in semantic space without a verbalized confidence interface. CSR combines the correctness reward with a novel semantic calibration reward that encourages exploitation among correct rollouts by promoting semantic agreement, and exploration among incorrect ones by discouraging spurious consistency. Experiments across three model families on HotpotQA (in-distribution) and TriviaQA, MSMARCO, and NQ-Open (out-of-distribution) show that CSR consistently achieves lower ECE and higher AUROC than verbalized-confidence baselines across nearly all settings, reducing ECE by up to $40\%$ and improving AUROC by up to $31\%$ over verbalized-confidence baselines, with calibration behavior generalizing robustly across all four evaluation settings.
Read the full abstract ▾
All authors Fengfei Yu, Ruijia Niu, Dongxia Wu, Yian Ma, Rose Yu
Machine Learning
Embracing Biased Transition Matrices for Complementary-Label Learning with Many Classes
Tan-Ha Mai, Chao-Kai Chiang, Han-Hwa Shih et al. (+3 more)
Complementary-label learning (CLL) is a weakly supervised paradigm where instances are labeled with classes they do not belong to. Despite a decade of research, CLL methods remain competitive mainly on 10-class classification, with scaling to large label spaces continuing to be an enduring bottleneck. This limitation stems from the common assumption of uniform label generation in traditional methods, which fatally dilutes the learning signal in many-class settings. In this paper, we demonstrate that this long-standing barrier can be overcome by deliberately designing a biased (non-uniform) generation process that restricts complementary labels to a subset of classes. This finding motivates us to propose Bias-Induced Constrained Labeling (BICL), a principled framework spanning data collection to training that leverages this bias. BICL enables effective learning on CIFAR-100 and TinyImageNet-200, achieving more than sevenfold accuracy improvements over traditional methods. Our findings establish a new trajectory for making CLL feasible for many classes in real-world applications.
Read the full abstract ▾
All authors Tan-Ha Mai, Chao-Kai Chiang, Han-Hwa Shih, Gang Niu, Masashi Sugiyama, Hsuan-Tien Lin
Machine Learning
Gaussian Relational Graph Transformer
Zezhong Ding, Jin Li, Xugang Wang et al. (+1 more)
Relational graph learning enables predictive modeling over relational databases by directly capturing dependencies across interconnected tables. While relational graph
Transformers extend the receptive field beyond local message passing, a larger receptive field does not necessarily provide more useful information: sampling must preserve relational structure without introducing excessive semantically irrelevant nodes, while
attention must account for the temporal relevance of the retained information. We propose GelGT, a Gaussian relational graph
transformer that explicitly addresses these challenges. GelGT introduces a structure-semantic collaborative sampling strategy to preserve structural connectivity while filtering irrelevant semantic information, and incorporates a Gaussian graph
attention mechanism with a learnable Gaussian bias on the sampled subgraphs to dynamically encode temporal dependencies. We provide theoretical analysis characterizing the structural preservation, semantic refinement, and temporal discrimination of these mechanisms. Experiments on 7 real-world relational
datasets covering 21 prediction tasks show that GelGT consistently outperforms existing relational graph learning methods, with improvements of up to 13.8%.
Read the full abstract ▾
All authors Zezhong Ding, Jin Li, Xugang Wang, Xike Xie
Machine Learning · Cross-listed from cs.CL
Response-Conditioned Parallel-to-Sequential Orchestration for Multi-Agent Systems
Nurbek Tastan, Alex Iacob, Lorenzo Sani et al. (+4 more)
Multi-
agent systems can solve complex tasks through collaboration between multiple Large
Language Model
agents. Existing collaboration frameworks typically operate in either a parallel or a sequential mode. In the parallel mode,
agents respond independently to queries followed by aggregation of responses. In contrast, sequential systems allow
agents to communicate via a directed
topology and refine one another step by step. However, both modes are inadequate for achieving the desired objectives of minimizing communication and latency while simultaneously maximizing the accuracy of the final response. In this work, we introduce a hybrid paradigm called Nexa, a trainable response-conditioned policy that bridges the gap between the two modes. Nexa begins with a parallel execution stage, embeds the resulting responses into a shared semantic space, and then predicts a sparse directed acyclic communication graph. If the graph is empty, the system remains purely parallel; if it is non-empty, the system performs one sequential message propagation. The policy is a lightweight
transformer model, and the method avoids the need for external LLM judges or reward models, as well as hand-crafted test-time
topology search. We formalize this hybrid execution problem, show that the resulting graph is acyclic by construction, and that the framework strictly subsumes pure parallel execution, and present a training procedure based on policy-gradient optimization. Results demonstrate that the response-conditioned policy learned by Nexa under one setting can be reused when the number of
agents, the task, or the underlying
agent changes, thus emphasizing the generalizability of the learned communication policy.
Read the full abstract ▾
All authors Nurbek Tastan, Alex Iacob, Lorenzo Sani, Meghdad Kurmanji, Nicholas D. Lane, Samuel Horvath, Karthik Nandakumar
Machine Learning · Cross-listed from stat.ML
MaxSketch: Robust Distinct Counting in Streams via Random Projections
Nikos Tsikouras, Constantine Caramanis and Christos Tzamos
Estimating the number of distinct elements in a data stream is well understood when repeated elements are identical. In modern settings, however, observations are high-dimensional and noisy, so repeated instances of the same object are only approximately similar – for example, different images of the same individual may vary significantly at the pixel level. Classical sketches such as HyperLogLog rely on consistent hash values for identical elements and break down in this regime. Recent work on robust distinct counting in general metric spaces achieves $\widetilde{\Theta}(\sqrt{n})$
memory, which is tight in the worst case. We show that substantially improved
memory guarantees are possible under
geometric structure common in learned representations. We introduce MaxSketch, a simple max-linear sketch built from random Gaussian projections, and prove that it succeeds in estimating the number of distinct latent objects. Concretely, we show that under this assumption $m = \widetilde{O} (\log n / \varepsilon^2)$ random projections (and hence $\widetilde{O} (\log n/\varepsilon^2)$
memory) suffice to recover the true distinct count within a $(1+\varepsilon)$ factor. Experiments on image streams confirm that MaxSketch accurately estimates distinct counts and generalizes beyond the training regime. Our results bridge classical streaming
algorithms and modern representation learning, showing how
geometric structure can fundamentally reduce the complexity of distinct counting.
Read the full abstract ▾
Machine Learning
AstraFlow: Dataflow-Oriented Reinforcement Learning for Agentic LLMs
Haizhong Zheng, Yizhuo Di, Jiahui Wang et al. (+7 more)
Reinforcement learning (RL) is increasingly used to improve the
reasoning, coding, and tool-use capabilities of large
language models, but agentic RL remains prohibitively expensive. Scaling RL to agentic LLMs requires supporting complex workloads, including multi-policy collaborative training, while efficiently using elastic, heterogeneous, and cross-region compute resources. Existing LLM RL systems support some of these capabilities, but each new extension often requires dedicated system engineering. This burden arises from trainer-centered control architectures and the lack of principled abstractions for RL system components. To address these limitations, we propose AstraFlow, a dataflow-oriented RL system that replaces conventional trainer-centered control with principled component abstractions. In AstraFlow, rollout services, dataflow management, and training are decoupled into autonomous components, enabling the system to natively support complex multi-policy agentic RL workloads and efficiently exploit diverse compute resources. We evaluate AstraFlow across math, code, search, and AgentBench workloads, showing that the same system supports multi-policy training, elastic scaling, heterogeneous cross-region execution, and composable data
algorithms without system-level code changes. In multi-policy collaborative training, AstraFlow achieves comparable or better accuracy than existing RL systems while speeding up training time by 2.7x.
Read the full abstract ▾
All authors Haizhong Zheng, Yizhuo Di, Jiahui Wang, Shuowei Jin, Xueshen Liu, Yongji Wu, Z. Morley Mao, Ion Stoica, Jiawei Zhao, Beidi Chen
Machine Learning
CrystalBoltz: End-to-End Protein Structure Determination via Experiment-Guided Diffusion for X-Ray Crystallography
Minseo Kim, Huanghao Mai, Jay Shenoy et al. (+3 more)
Generative models trained on public databases of
protein structures, most of which have been determined by X-ray crystallography, now provide powerful priors for structure prediction. However, they are not readily conditioned on the measurements from a new crystallographic experiment, limiting their use for X-ray structure determination. In crystallography, the measured structure-factor amplitudes do not by themselves determine an electron density map or atomic structure because the associated phases are unobserved and must be inferred. Structure determination therefore remains an inverse problem in which candidate models must be both structurally plausible and consistent with measured diffraction data, often requiring substantial manual refinement by human experts. Emerging methods aim to incorporate experimental information more directly into predictive and refinement workflows. We present CrystalBoltz, a generative framework that casts crystallographic refinement as
Bayesian inference over atomic structures and operates directly on structure-factor amplitudes. CrystalBoltz moves from unguided generation with a pre-trained prior over
protein structures to experiment-guided posterior sampling, followed by atomic coordinate and B-factor refinement. Across multiple
protein crystallography
datasets, CrystalBoltz attains lower coordinate RMSD and lower R-factors than the strongest baselines considered, while reducing runtime by a factor of 33 relative to existing experimentally guided refinement.
Read the full abstract ▾
All authors Minseo Kim, Huanghao Mai, Jay Shenoy, Alec Follmer, Gordon Wetzstein, Frederic Poitevin
Machine Learning · Cross-listed from cs.CL
When Latent Geometry Is Not Enough: Draft-Conditioned Latent Refinement for Non-Autoregressive Text Generation
De Shuai Zhang
Continuous
diffusion and flow models are attractive for non-autoregressive text generation because they can update all positions in parallel. A major difficulty is the interface between continuous latent states and discrete
tokens. This report studies a draft-conditioned latent refinement model built from a frozen BERT encoder, a parallel decoder, a denoising DraftPrior, a local FlowNet, and a learned diagonal MetricNet. Early Gaussian-start experiments showed that good latent-space metrics, such as scale matching or cosine similarity, do not guarantee good decoding. Generated latents can be close to real encoder latents but still produce high-
entropy, biased, or repetitive
token distributions. We therefore frame the task as controlled local refinement rather than full generation from noise. On ROCStories, using the first two sentences as
prompt and the last three as target, full 768-dimensional BERT latents recover
tokens much better than compressed 256-dimensional latents. With 768-dimensional latents, DraftPrior target-
token probability is 0.938 for clean drafts, 0.613 for 3%
token dropout, 0.483 for 5% dropout, and 0.272 for 10% dropout. Local flow refinement and fused decoder-aware readout give modest additional gains, while metric learning and OT-style
alignment improve geometry but do not close the decoder gap. The main result is a diagnostic one: latent geometry alone is not enough. Continuous latent text generation should be evaluated by decoder recoverability, the quality of the start distribution, and whether refinement preserves decoder-readable structure.
Read the full abstract ▾
Machine Learning
Characterizing Learning in Deep Neural Networks using Tractable Algorithmic Complexity Analysis
Pedram Bakhtiarifard, Sophia N. Wilson, Mahmoud H. A. Afifi et al. (+2 more)
Training large-scale deep neural networks (DNNs) is resource-intensive, making model compression a practical necessity. The widely accepted ”learning as compression” hypothesis posits that training induces structure in network weights, which enables compression. Measuring this structure through Kolmogorov-Chaitin-Solomonoff (KCS) complexity is appealing, but existing
estimators based on the Coding
Theorem Method (CTM) and the Block Decomposition Method (BDM) are limited to small binary objects and do not scale to modern DNNs. We introduce the Quantized Block Decomposition method (QuBD), which extends algorithmic complexity estimation to any $k$-ary object. QuBD first quantizes the network weights to a finite alphabet, then estimates the KCS complexity by aggregating per bit-plane CTM estimates. We show theoretically that QuBD yields a tighter upper-bound on the estimation gap for ideal prefix KCS complexity under bit-plane independence than binarization-based methods. Using QuBD, we study how the algorithmic complexity of neural network weights evolves during training, showing that it decreases as models learn, scales with data budget, increases during overfitting, follows the delayed generalization observed during grokking, and correlates with generalization performance. We further show that algorithmic information resides predominantly in the most significant bit-planes, which can serve as a practical diagnostic for determining appropriate post-training quantization levels. This work offers novel insights into learning mechanisms in DNNs by providing the first scalable, tractable estimates of KCS complexity for large, non-binary objects such as DNN weights.
Read the full abstract ▾
All authors Pedram Bakhtiarifard, Sophia N. Wilson, Mahmoud H. A. Afifi, Jonathan Wenshøj, Raghavendra Selvan
Machine Learning
CTF4Nuclear: Common Task Framework for Nuclear Fission and Fusion Models
Stefano Riva, Carolina Introini, Antonio Cammi et al. (+13 more)
The demand for clean energy is ever increasing, with new
nuclear technologies presenting a complementary solution to renewable energies. However, designing and operating these systems is exceptionally difficult, given the complexity of the physical phenomena that interact to form the system dynamics. While high-fidelity simulations help to understand the non-linear, multi-physics interactions within a reactor, they are computationally expensive and rarely suitable for real-time applications. Furthermore, model-based approaches are inherently sensitive to simplifying assumptions required to derive their governing equations and parameters, leading to inevitable discrepancies with real-world measurements. In contrast, Machine Learning (ML) methods have the potential to generate reliable surrogate models which may be able to quickly predict the system's behaviour. However, the number of data-driven methods that can potentially be used for this task is large and diverse. In a safety-critical setting such as
nuclear engineering, a fair comparison of different ML methods, and a clear understanding of their advantages and limitations, is of paramount importance. To address this, we introduce a Common Task Framework (CTF) for ML in
nuclear engineering, building upon previous efforts in dynamical systems and seismology. This CTF considers a curated set of
datasets from different
nuclear and
nuclear-adjacent systems. The CTF evaluates the performance of a method on 12 established metrics, alongside a new paradigm focused on system monitoring from sparse measurements only. We illustrate the framework by benchmarking standard ML baselines against these
datasets, revealing current method limitations. Our
vision is to replace ad hoc comparisons with standardized evaluations on hidden test sets, raising the bar for rigour and reproducibility in scientific ML for the
nuclear industry.
Read the full abstract ▾
All authors Stefano Riva, Carolina Introini, Antonio Cammi, Dean Price, Alexey Yermakov, Yue Zhao, Philippe M. Wyder, Judah Goldfeder, Jan Williams, Amy Sara Rude, Matteo Tomasetto, Joe Germany, Joseph Bakarji, Georg Maierhofer, Miles Cranmer, J. Nathan Kutz
Machine Learning
DeltaPrompts: Escaping the Zero-Delta Trap in Multimodal Distillation
Jaehun Jung, Hyunwoo Kim, Brandon Cui et al. (+4 more)
Distillation enables compact
Vision-
Language Models (VLMs) to obtain strong
reasoning capabilities, yet the
prompts driving this process are typically chosen via simple heuristics or aggregated from off-the-shelf
datasets. We reveal a critical inefficiency in this approach: up to 69% of the
prompts in standard chart / document
reasoning datasets are effectively zero-delta, meaning the teacher and student already induce the exact same answer distribution. Training on these
prompts provides minimal learning signal, causing student improvement to rapidly saturate regardless of data scale. To escape the zero-delta trap, we return to first principles:
distillation fundamentally minimizes distributional divergence, and thus a
prompt is valuable only if it exposes a functional capability gap between the teacher and student. We quantify this gap through answer divergence ($\Delta$), demonstrating that non-zero divergence is critical for effective scaling. Building on this insight, we propose a staged synthesis pipeline that repurposes existing
datasets as seeds, actively targeting student failure modes to produce better
prompts. The result is DeltaPrompts, a diverse
dataset of 200k synthetic, high-divergence
reasoning problems. We evaluate DeltaPrompts across three distinct settings: on-policy
distillation with the target teacher-student pair, transfer to a novel model family without regenerating the data, and off-policy fine-tuning of a non-
reasoning model. Across all scenarios, DeltaPrompts drives substantial gains, yielding up to 15% relative improvement even on top of a highly-optimized
reasoning model (e.g., Qwen3-VL-8B-Thinking) – averaged over 10
benchmarks spanning chart, document and perception-centric
reasoning.
Read the full abstract ▾
All authors Jaehun Jung, Hyunwoo Kim, Brandon Cui, Ximing Lu, David Acuna, Prithviraj Ammanabrolu, Yejin Choi
Machine Learning
Rethinking Neural Network Learning Rates: A Stackelberg Perspective
Sihan Zeng, Sujay Bhatt and Sumitra Ganesh
Neural networks are typically trained with a single learning rate across all layers. While recent empirical evidence suggests that assigning layer-specific learning rates can accelerate training, a principled understanding of the conditions and mechanisms under which non-uniform learning rates are beneficial remains limited. In this work, we investigate non-uniform learning rates through the lens of Stackelberg optimization. Specifically, we demonstrate that training neural networks with a smaller learning rate for the body layers and a larger learning rate for the final layer can be interpreted as a two-time-scale alternating gradient descent
algorithm applied to a Stackelberg reformulation of the original objective. We establish finite-time convergence guarantees for the
algorithm under broad conditions that accommodate constraint sets and non-smooth activation functions. Beyond convergence, we identify two mechanisms by which non-uniform learning rates can outperform uniform learning rates: (i) we show that certain problem instances induce a Stackelberg objective with stronger optimization structure than the original objective, yielding faster convergence to globally optimal solutions, (ii) our numerical analysis reveals that the Stackelberg objective can exhibit substantially sharper local
curvature, especially in early training, which leads to more informative gradients and learning acceleration. Experiments in supervised learning and reinforcement learning support our findings.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.CL
Process Rewards with Learned Reliability
Jinyuan Li, Langlin Huang, Chengsong Huang et al. (+5 more)
Process Reward Models (PRMs) provide step-level feedback for
reasoning, but current PRMs usually output only a single reward score for each step. Downstream methods must therefore treat imperfect step-level reward predictions as reliable decision signals, with no indication of when these predictions should be trusted. We propose BetaPRM, a distributional PRM that predicts both a step-level success probability and the reliability of that prediction. Given step-success supervision from Monte Carlo continuations, BetaPRM learns a Beta belief that explains the observed number of successful continuations through a Beta-Binomial likelihood, rather than regressing to the finite-sample success ratio as a point target. This learned reliability signal indicates when a step reward should be trusted, enabling downstream applications to distinguish reliable rewards from uncertain ones. As one application, we introduce Adaptive Computation Allocation (ACA) for PRM-guided Best-of-N
reasoning. ACA uses the learned reliability signal to stop when a high-reward solution is reliable and to spend additional computation on uncertain candidate prefixes. Experiments across four backbones and four
reasoning benchmarks show that BetaPRM improves PRM-guided Best-of-N selection while preserving standard step-level error detection. Built on this signal, ACA improves the accuracy–
token tradeoff over fixed-budget Best-of-16, reducing
token usage by up to 33.57% while improving final-answer accuracy.
Read the full abstract ▾
All authors Jinyuan Li, Langlin Huang, Chengsong Huang, Shaoyang Xu, Donghong Cai, Yuyi Yang, Wenxuan Zhang, Jiaxin Huang
Machine Learning
Neural Point-Forms
Bruno Trentini, Jacob Hume, Vincenzo Antonio Isoldi et al. (+3 more)
Point cloud learning often rests on the premise that observed samples are noisy traces of an underlying
geometric object, such as a
manifold embedded in a high-dimensional feature space. Yet much of this geometry is not captured directly by coordinates, pairwise distances, or learned graph neighborhoods alone. In the smooth setting, differential forms are devices to encode higher order tangency information. In this work, we introduce a new family of principled learnable
geometric features for point clouds called neural point-forms (NPFs). In the absence of a natural tangency structure, we instead use
Laplacian-based techniques from
Diffusion Geometry to build a discrete model for comparing differential forms on point clouds via inner products. In the continuum, submanifolds of a shared ambient feature space are represented as comparison
matrices, whose entries describe how pairs of feature forms interact with extrinsic tangency information. We make this intuition precise by proving the long-run consistency of comparison
matrices under standard sampling, bandwidth, density, and
manifold-hypothesis assumptions. This yields a compact, efficient and permutation-invariant neural layer whose output is a learned form-comparison
matrix. Across synthetic and biologically relevant experiments, we show that NPFs provide a competitive, and interpretable representation, with the strongest benefits appearing when labels depend on sampling density,
manifold-like structure, or response-relevant population geometry.
Read the full abstract ▾
All authors Bruno Trentini, Jacob Hume, Vincenzo Antonio Isoldi, Philipp Misof, Ekaterina S. Ivshina, Kelly Maggs
Machine Learning · Cross-listed from math.OC
Stochastic Non-Smooth Convex Optimization with Unbounded Gradients
Dmitry Kovalev
Much of the existing theory on first-order non-smooth optimization is built on a restrictive assumption that the gradients of the objective function are uniformly bounded. We introduce a much more realistic class of generalized Lipschitz functions, where the gradient norms are bounded by an affine function of the optimality gap. We then ask a natural question: what
algorithm achieves the best global convergence rates for solving convex
stochastic generalized Lipschitz optimization problems? To address this, we develop a new convergence analysis for several existing
algorithms and find that AdamW with clipped updates, provably outperforms other popular
stochastic optimization methods, such as SGD and AdaGrad. Moreover, our analysis establishes the critical role of AdamW's exponentially weighted gradient accumulation, as opposed to simple averaging. We further show that clipped AdamW is universal and achieves improved rates under the popular generalized smoothness assumption, analyze the convergence of clipped AdamW with diagonal and
matrix preconditioners, and extend our results to the
quasar-convex setting.
Read the full abstract ▾
Machine Learning
On the Fragility of Data Attribution When Learning Is Distributed
Xian Gao, Bo Hui, Min-Te Sun et al. (+1 more)
Data attribution has become an important component of pricing,
auditing, and governance in machine learning pipelines, yet most attribution methods implicitly assume that attribution values faithfully reflect participants' contributions. We show that this assumption can fail: a single participant in a standard distributed training workflow can substantially inflate its measured attribution value while preserving global utility. Our attribution-first attack uses latent optimization to inject small synthetic batches that preserve utility while exploiting non-IID label coverage and evaluator sensitivities. Across
datasets, models, and multiple marginal-utility evaluators, the attack consistently increases the adversary's attribution value and reshapes the relative attribution structure among benign clients without degrading accuracy or triggering geometry-based defenses. These results show that attribution itself forms a new attack surface and motivate the development of attribution-robust and incentive-compatible scoring mechanisms.
Read the full abstract ▾
All authors Xian Gao, Bo Hui, Min-Te Sun, Wei-Shinn Ku
Machine Learning · Cross-listed from cs.CL
RoPE Distinguishes Neither Positions Nor Tokens in Long Contexts, Provably
Yufeng Du, Phillip Harris, Minyang Tian et al. (+5 more)
We identify intrinsic limitations of Rotary Positional
Embeddings (RoPE) in
Transformer-based long-context
language models. Our theoretical analysis abstracts away from the specific content of the context and depends only on its length. We prove that as context length increases, RoPE-based
attention becomes unpredictable and loses two properties that are central to its effectiveness. First, it loses its locality bias: RoPE is no more likely to favor nearer positions than substantially farther ones. Second, it loses consistency in
token relevance: a key vector that receives a higher
attention score than an alternative at one position may receive a lower score at another. In both cases, the probability of failure approaches 0.5, no better than random guessing. We further prove that the
attention score can remain unchanged when a key
token is moved to a different position, or even replaced by a different
token, indicating a failure to distinguish positions or
tokens. Adjusting the RoPE base trades off distinguishing positions against distinguishing
tokens but cannot preserve both at the same time. Increasing the RoPE base hyperparameter, a common practice in today's long-context models, helps distinguish different
tokens, but inevitably sacrifices the ability to distinguish positions. Our empirical analysis shows that multi-head, multi-layer architectures are insufficient to overcome these limitations. Our findings suggest that fundamentally new mechanisms for encoding position and
token order may be needed in future
Transformer long-context
language models.
Read the full abstract ▾
All authors Yufeng Du, Phillip Harris, Minyang Tian, Eliu A Huerta, Srikanth Ronanki, Subendhu Rongali, Aram Galstyan, Hao Peng
Machine Learning
OgBench: A Framework for Evaluating Graph Neural Networks on Omics Data
Louisa Cornelis, Johan Mathe, Louis Van Langendonck et al. (+2 more)
Graph Neural Networks (GNNs) have become the dominant framework for inductive graph-level learning. Yet most
benchmarks focus on the regime $n \gg p$, where the number of graphs $n$ greatly exceeds the number of nodes per graph $p$. This overlooks biological domains such as omics, which operate in the opposite $n \ll p$ regime, characterized by large graphs of genes, transcripts, or
proteins across few patient samples. This raises the question: how do GNNs perform in this low-sample, high-node omics setting? We introduce OgBench (Omics-Graph Bench), the first benchmarking platform for graph-level prediction in the $n \ll p$ regime characteristic of omics data. We provide a standardized, end-to-end modular infrastructure from raw omics data to families of featured graphs with varied structural properties. We
benchmark classical GNNs, as well as GNNs designed for large graphs and omics applications, alongside MLPs and machine learning baselines to establish reference performances. Our results show that widely used GNNs often do not outperform simple MLPs and classical baselines. These findings challenge the prevailing assumption that graph structure inherently adds value in this domain, fostering a critical reassessment of current learning paradigms. Ultimately, by exposing these limitations, OgBench provides the open-source ecosystem necessary for the community to develop and validate novel architectures explicitly tailored for biological graphs. The code is available at https://github.com/
geometric-intelligence/ogbench.
Read the full abstract ▾
All authors Louisa Cornelis, Johan Mathe, Louis Van Langendonck, Guillermo Bernárdez, Nina Miolane
Machine Learning
parallelcbf: A composable safety-filter and auditability framework for tensor-parallel reinforcement learning
Yijun Lu, Zilei Yang and Yuyin Ma
While Isaac Lab provides massive parallel
UAV simulation, OmniSafe and safe-control-gym provide constrained-RL
benchmarks, and CBFKit provides control-barrier-function synthesis tooling, no existing framework unifies these capabilities for end-to-end safety-constrained training. ParallelCBF is the first framework to unify (i)~tensor-parallel
UAV environments, (ii)~hard-gate CBF safety filters, (iii)~sharded BC-to-RL pipelines, and (iv)~first-class operational auditability – pre-registration, watchdog registries, failure forensics, and
dataset audits as composable APIs rather than user-implemented scripts. We release ParallelCBF v0.1.0 under Apache~2.0 with a four-layer composable API, a CPU PyTorch reference implementation of a dual-barrier (squared / linear-predictive) CBF, property-based safety invariance tests across vectorized batch sizes that complete in 1.67~s for the full 39-test suite, and a 31{,}415-episode behavior-cloning collection campaign whose curriculum mix, per-bucket yields, and
dataset SHA-256 are auditable through the framework's own ops primitives. We report a representative end-to-end pipeline execution in which the framework's auditability layer halted a downstream training stage that did not meet pre-registered convergence criteria, preventing silent propagation of a degraded checkpoint – an architectural property we argue is necessary, not merely useful, for reproducible empirical robotics research. The framework is installable via pip install parallelcbf; source and release artifacts are available at https://github.com/xiaoyang-123-
cell/ParallelCBF.
Read the full abstract ▾
Machine Learning
STS: Efficient Sparse Attention with Speculative Token Sparsity
Jiangnan Yu, Ceyu Xu, Yongji Wu et al. (+1 more)
The quadratic complexity of
attention imposes severe
memory and computational bottlenecks on Large
Language Model (LLM) inference. This challenge is particularly acute for emerging agentic applications that require processing multi-million
token sequences. We propose STS, a sparse
attention mechanism that requires no model retraining. STS leverages the key insight that
tokens identified as important by a smaller draft model are highly predictive of important
tokens for a larger target model. By integrating into speculative decoding frameworks, STS repurposes the draft model's
attention scores to dynamically construct a
token-and-head-wise sparsity mask. This mask effectively prunes the expensive
attention computation in the target LLM. Our evaluation shows that STS achieves a 2.67x speedup operating at approximately 90% sparsity on representative
benchmark NarrativeQA, maintaining negligible accuracy degradation compared to dense
attention. STS establishes a new state-of-the-art on the sparsity-accuracy trade-off, outperforming prior techniques by enabling higher sparsity levels for a given accuracy budget.
Read the full abstract ▾
All authors Jiangnan Yu, Ceyu Xu, Yongji Wu, Yuan Xie
Machine Learning · Cross-listed from cs.IT
PrismQuant: Rate-Distortion-Optimal Vector Quantization for Gaussian-Mixture Sources
Bumsu Park, Chanho Park, Youngmok Park et al. (+1 more)
For a Gaussian source under mean-squared error (MSE), classical transform coding is rate–distortion (RD) optimal: the Karhunen–Loeve transform (KLT) diagonalizes the covariance, reverse waterfilling allocates the bits, and scalar quantization closes the loop. This elegant story breaks down for
multimodal sources, where no single covariance can capture heterogeneous local geometries, and the RD function loses its closed form. We revisit this problem through Gaussian-mixture sources and develop a constructive RD theory for them. Our key finding is that the mixture structure incurs only a component label cost. Conditioned on the active mixture component, each branch is Gaussian; the challenge is allocating bits across heterogeneous branches. We prove that the genie-aided conditional RD function is governed by a single global reverse-waterfilling level shared across all components and eigenmodes. Building on this result, we introduce PrismQuant, which transmits the component label losslessly and encodes the residual using the component-matched KLT, followed by scalar quantization, achieving a rate of H(C)/n bits per source dimension of the converse, with a vanishing asymptotic gap. We further develop a practical implementation based on EM-driven Gaussian-mixture learning, component-adaptive KLTs, and
entropy-constrained scalar quantization (ECSQ). Experiments on synthetic Gaussian mixtures show that PrismQuant closely approaches the theoretical RD bound, while experiments on real-world channel-state-information (CSI) data demonstrate competitive or superior performance compared with
transformer-based learned codecs at more than one order of magnitude smaller model size.
Read the full abstract ▾
All authors Bumsu Park, Chanho Park, Youngmok Park, Namyoon Lee
Machine Learning · Cross-listed from cs.AI
X-SYNTH: Beyond Retrieval – Enterprise Context Synthesis from Observed Digital Human Attention
Guruprasad Raghavan, George Nychis and Rohan Narayana Murthy
In enterprise operations, the context required for an AI
agent task is scattered across systems of record, static information stores, and communication channels. What is stored is system state, a lossy representation of the work that actually happened. The prevailing approach retrieves by matching request content to what is stored; for narrow requests this works well. But synthesis quality depends on knowing what to surface and how to interpret it: knowledge specific to each organization, team, and individual, present in behavioral patterns, absent from any
retrieval index. For the agentic task of proposing enterprise-valuable leads to sellers, this approach breaks down: True Lead Rate is low, False Lead Rate is high, and the model has no mechanism to improve. We present X-SYNTH, a framework for enterprise context synthesis grounded in digital human
attention, the digitally observable interaction signatures of each worker, encoding what they did, the sequence in which they did it, and implicit reward signals. Behavioral traces preceding positive outcomes are distinguishable from those that did not, without external labeling. X-SYNTH models each individual's behavioral baseline as a Digital Twin Signature (DTS) and selects among seven
attention filters, Proportional, Inverse, Differential, Recurrent, Comparative, Sequential, and Collective, per individual and per query, to identify causally relevant activity signatures. A four-stage pipeline assembles ranked context grounded in behavioral patterns rather than query
embeddings. A frontier model unaided achieves 9.5% True Lead Rate (TLR) with 90.5% False Lead Rate (FLR). Augmented with X-SYNTH, TLR rises to 61.9% (6.5x) while FLR falls to 18.8%. Enterprise context synthesis is not a
retrieval problem. It is a relevance problem, and digital human
attention is its most reliable ground truth.
Read the full abstract ▾