Machine Learning · Cross-listed from math.OC
Scalable Incremental Robustness Analysis of Neural Network Feedback Systems
Zichen Wang, Peter Seiler, Geir Dullerud et al. (+1 more)
Semidefinite programming (SDP) certificates for feedback systems containing deep neural networks (NNs) typically scale with the total number of
neurons, whereas small-gain tests are scalable but can be highly conservative. This paper develops a unified and scalable framework for incremental robust stability and performance analysis of feedback interconnections involving high-dimensional NNs and unmodeled dynamics. By combining a structured decomposition of the full-order SDP condition with scalable Lipschitz constant estimation
algorithms, we derive reduced verification conditions that certify incremental convergence and incremental $\ell_2$-gain bounds. The dimensions of the resulting control-analysis linear
matrix inequalities (LMIs) depend only on the widths of the last two network layers and are independent of network depth. The framework preserves the coupling between the plant and the NN, with the incremental small-gain condition recovered as a special case. To further reduce conservatism, we develop a multi-round alternating update scheme that iteratively refines the coupling variables while preserving scalability. Numerical experiments show that the proposed framework achieves state-of-the-art incremental $\ell_2$-gain bounds for large-scale NN feedback systems.
Read the full abstract ▾
All authors Zichen Wang, Peter Seiler, Geir Dullerud, Bin Hu
Machine Learning · Cross-listed from cs.CL
Do Student LLMs Inherit OOD Robustness? Invariance-Weighted Distillation for Reliable Knowledge Transfer
Dileesha Kannangara and Sanghamitra Dutta
Knowledge
distillation (KD) aims to compress high-performance teacher LLMs into lightweight students. However, distilled students often exhibit substantial performance degradation in out-of-distribution (OOD) settings, a critical gap that remains underexplored. We identify two compounding mechanisms causing OOD performance degradation: (1) data spuriousness: students can learn spurious correlations in the
distillation dataset over genuine causal relationships; and (2) teacher capability: standard KD treats all samples uniformly, ignoring whether the teacher is guided by causal features or misled by spurious shortcuts on a given sample. To address these challenges, we propose Invariance-Weighted
Distillation (IWD), a theoretically grounded framework that dynamically reweights training samples using an estimate of the teacher's causal reliance derived from prediction invariance across multiple synthetic environments. IWD perturbs spurious cues while preserving core semantics, assigning higher
distillation weights to samples whose teacher predictions remain invariant, indicating greater reliance on causal rather than spurious features. We theoretically show that IWD reduces the student's Spurious-to-Causal (S2C) gradient ratio compared to standard uniformly weighted KD, driving the student toward more invariant representations. Experiments on four NLP
benchmarks (MNLI, SQuAD-v2, CoNLL-2003 NER, and SST-2) across two model families (DeBERTa-v3 and Qwen-2.5) demonstrate that IWD consistently outperforms strong KD baselines on OOD evaluations while maintaining competitive in-distribution (ID) performance. Specifically, IWD achieves the highest accuracy in 15 out of 16 OOD
benchmarks and improves average OOD performance over standard KD by 4.34 percentage points on NLI and 14.94 percentage points on QA.
Read the full abstract ▾
Machine Learning
The Ups and Downs of Backprop Weights
Giuseppe Chindemi and Benjamin F. Grewe
Backpropagation (BP) has driven the remarkable success of modern deep learning by enabling large hierarchical networks to learn complex functions end-to-end. Yet it does not by itself determine how parameters should be organized so that functional components can be reused and adapted selectively. For example, object recognition and motion prediction may depend on overlapping parameter sets, making them difficult to isolate or modify independently. We call this condition weight
entanglement. Modern architectures dynamically select which parts of a network process each sample: nonlinearities gate units,
attention selects interactions, and Mixture-of-Experts architectures route inputs to modules. Yet such selection does not ensure that the same functional component remains linked to an identifiable parameter set across samples. We propose weight operators: parameterized modules that implement reusable functional components and can be composed at inference to form the function required by each sample. Learning proceeds in two stages: the model first infers the required operator composition, then updates only the selected operators' parameter sets. Vector Networks (VNs) provide one implementation. They couple operator selection to local error-driven updates within each layer and show that learned operators can be reused in combinations absent from training while updates remain restricted to the selected parameter sets. This provides a basis for testing functional parameter identifiability: whether an operator remains linked to the same functional component during learning. We argue that functional parameter identifiability may provide an organizing principle for models that systematically reuse and recombine learned functions while adapting only the components that need to change.
Read the full abstract ▾
Machine Learning · Cross-listed from stat.ML
RLVR is a Kernel, Not a Function: Statistical Inference for pass@$k$ Crossovers
Chen Yang and Jun Chen
Reinforcement learning with verifiable rewards (RLVR) often improves pass@1 while falling behind its base model at larger sampling budgets $k$, a crossover read as evidence that RLVR only sharpens existing capability. We identify two limits to this reading. First, a visible crossing need not be statistically established: comparing models on the same
prompts, we build confidence bands across sampling budgets $k$ that require evidence of both an early gain and a later loss. Across five public RLVR pairs no crossing is statistically established in the initial evaluations, while a 32k-
token evaluation on fresh
prompts locates a reversal with first loss between 11 and 61 samples; power analysis shows why failure to detect a crossing need not mean no crossing, and why more
prompts can help more than more answers per
prompt. Second, base success alone does not determine what RLVR does to a
prompt:
prompts with the same base success rate have different post-RL success rates, and these differences repeat across independent generation halves. The relationship is a conditional distribution—a Markov kernel—rather than a single curve, and fitting it predicts crossings in independent generations for the same
prompts and corrects the simple model's power estimates. Theory further shows how losses on a minority of the hardest
prompts can overturn an early lead even when training improves other
prompts, separating evidence that a crossover exists from claims about what it means for capability.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.GT
Strategic Classification Has a Missing Lever: Audit Risk
Raman Ebrahimi and Massimo Franceschetti
Strategic classification studies how a decision maker should choose a classifier when the
agents being classified can adjust their features in response to it. In existing models, the classifier is the only instrument available to the decision maker, and therefore a feature that is predictive but easy to fake can only be down-weighted or discarded. However, in many settings the decision maker can also verify: lenders verify income, admissions offices check documents, and tax authorities audit returns. In this paper, we propose a model of strategic classification in which the firm jointly designs a linear classifier and an audit profile, which assigns to each fakeable feature a probability of detection and a penalty when caught. We show that under linear costs, the classifier affects the audit problem only through the distribution of gaming rents it induces, so that the joint design problem decomposes into the choice of a score rule and an audit allocation problem. We use this decomposition to characterize the optimal audit allocation, to identify when the allocation problem is tractable and when it is NP-hard (namely, when
agents can game through overlapping features under an inspection cap), and to bound the regret of a firm that has to learn the rents by
auditing. We further show that audit intensity is a quantity to be tuned rather than maximized: welfare is single-peaked in it, and a firm and a social planner disagree on the mix of detection and penalty that delivers a given level of deterrence. Notably, two populations with identical costs and causal structure can game in one domain and improve in the other, a difference that a cost-only model cannot account for. Together, our findings highlight that whether a feature is “gameable” depends on the institution's verification policy as much as on the feature itself.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.CR
Defusing Explosive Prompts: Understanding and Preventing Trigger-Based Prompt Injections in LLM Agents
Justin Szczepaniak, Elad Feldman, Naum Viner et al. (+1 more)
As LLM applications integrate with external tools, they are increasingly exposed to indirect
prompt injection (IPI), where
adversarial instructions are embedded in retrieved content. Conventional IPIs fire on contact: the moment an
agent ingests the content, it carries out the instruction. We introduce the explosive
prompt, a conditional payload that stays dormant until an attacker-chosen trigger is met, in effect a training-free, inference-time
backdoor planted in a single piece of retrieved content. This temporal separation reaches where ordinary IPI cannot. On frontier models that refuse the bare imperative almost entirely, rephrasing the same goal as a dormant conditional drives real, state-changing tool execution against a live
agent backend (a paired mean of 16.5% vs. 2.4% for the imperative, reaching 34.2% on a proprietary model). In trials on nine production
agents (OpenAI Codex, Google Gemini CLI, Anthropic Claude Code CLI, Cursor CLI, GitHub Copilot, Devin AI CLI, Amazon Kiro CLI, Qwen Code, Google Assistant; n=30 each), explosive
prompts succeed in 43-83% of cases versus at most 3% for an imperative baseline, and they slip past deployed defenses: off-the-shelf injection classifiers are miscalibrated on them, and a preference-optimized model that closes imperative injection entirely still executes 11.8% of explosive
prompts, every one at the trigger turn. The durable defensive lever is ingestion-time detection of the conditional structure, once detectors are trained on explosive-
prompt data, which no prior
benchmark supplied and our generator does. Retraining cuts live tool-execution attack success from an undefended 34.3% to 7.5-8.1% for the encoder baselines. Our detector, DeFuse, reaches 3.0% at a calibrated 5% false-positive budget with the best detection quality of any method tested (AUC 0.9994) and 25x lower latency, though it needs length-aware thresholds.
Read the full abstract ▾
All authors Justin Szczepaniak, Elad Feldman, Naum Viner, Ben Nassi
Machine Learning
EmbeddGAN: A Novel GAN Framework Using an Embedding Network and Gini Distance Correlation
MaTais Caldwell, Yixin Chen, Xin Dang et al. (+1 more)
Generative
Adversarial Networks (GANs) have demonstrated strong performance in generating high-quality synthetic data. However, they are limited by no formal guarantees regarding convergence and the effectiveness of the learning process. In practice, this leads to training instability, mode collapse, and sensitivity to hyperparameters. To address this, we propose EmbeddGAN, a novel
adversarial training framework based on a dependence-based objective. Instead of relying on a discriminator that classifies samples as real or fake, EmbeddGAN introduces an
embedding network that learns a representation in which statistical dependence between samples and their real/fake labels is maximized, while the generator is trained to minimize this dependence. This objective is implemented using the Gini distance correlation (gCor), which equals zero if and only if the
embeddings are statistically independent of the real/fake label. Minimizing this objective therefore encourages real and generated samples to become statistically indistinguishable in the learned
embedding space. The
embedding network projects both real and generated data into a shared low-dimensional space, where distributional discrepancies can be measured directly through pairwise distances. We adopt a minimax training strategy: the
embedding network maximizes the Gini distance correlation (maximizing dependence), while the generator minimizes it (minimizing dependence). Experiments on the MNIST, CIFAR-10, and CelebA
datasets demonstrate that EmbeddGAN achieves competitive performance relative to established baselines while exhibiting notably stable training dynamics on the evaluated
datasets.
Read the full abstract ▾
All authors MaTais Caldwell, Yixin Chen, Xin Dang, Charles Walter
Machine Learning
COREM: Cosine-Relation Momentum Reshaping with Stateful Writeback
Yan Wang, Xiaochuan Wang and Yuxiang Sun
Matrix-valued optimizer states may contain relational structure that is not captured by treating their entries independently. We study whether relations within
matrix-valued optimizer states can be exploited to improve optimization. To this end, we introduce a unit-relation-transform abstraction and instantiate it as COREM, a Cosine-Relation Momentum Reshaping method with stateful writeback. COREM partitions the momentum state into update units, computes cosine relations among them, and uses these relations to reshape the momentum before writing the transformed state back to the optimizer. This stateful mechanism allows the reshaped momentum to affect not only the current update but also future optimization dynamics. We evaluate COREM on CIFAR-10 with an MLP and on enwik8 with a
Transformer. Compared with
Muon, COREM shows lower early-stage step efficiency but stronger improvement in the mid-to-late stages of training, achieving better final validation performance on CIFAR-10 and comparable final performance on enwik8. Spectral diagnostics on enwik8 show that COREM consistently increases
entropy effective rank and reduces the concentration of singular energy in dominant modes, while preserving an anisotropic spectrum. For square
matrix updates, COREM requires approximately 13.3% of the transformation FLOPs of
Muon with five Newton-Schulz iterations.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.NI
Semantics Delivery Network: Rethinking Web Retrieval Infrastructure for LLM Agents
Peichun Hua and Yunming Xiao
Large
language models (LLMs) increasingly rely on external sources when answering questions that require proprietary information or up-to-date live web content, through both traditional single-shot
retrieval-augmented generation (RAG) and multi-turn agentic RAG. Yet today's web infrastructure is still built for human clients. Given a query, current search services return a list of URLs and snippets ranked for generic relevance; content delivery networks (CDNs) cache URL-addressed objects (texts, images,
videos, etc.) without knowing which passage an
agent needs. LLMs, in contrast, consume short, semantically coherent passages, hereafter "chunks", selected for downstream task utility rather than similarity alone, and may retrieve statefully across
reasoning turns. Uncoordinated
agents also repeat search, data acquisition, and semantic processing, duplicating work that could be shared. We argue that semantic chunk
retrieval should become a first-class network-delivery abstraction. We propose Semantics Delivery Network (SemDN): an origin-authorized, hierarchical edge substrate that indexes, searches, and smart-caches web content at chunk granularity. SemDN serves
agents on behalf of participating websites, amortizes data acquisition and processing across
agents, and supports tenant-specific
retrieval policies. Because, unlike URL caching, semantic
retrieval provides no explicit miss signal, SemDN must estimate when its enrolled corpus may be incomplete or stale and trigger scoped discovery or refresh. It raises open questions about shareable
retrieval state, hierarchical caching, coverage risk, and deployment. Our preliminary probes reveal a large gap between page content processed and chunks consumed, substantial task-local reuse, and higher answer quality per context
token from chunk delivery.
Read the full abstract ▾
Machine Learning · Cross-listed from hep-ph
Machine Learning for Invisible Dark Boson Searches at the Electron-Ion Collider
Rojae Mighty and Ankush Reddy Kanuganti
We investigate whether adding $t$, the positive magnitude of the squared
nuclear four-momentum transfer, enables boosted decision trees (BDTs) to improve invisible-dark-
boson selection relative to optimized rectangular cuts at the Electron-Ion
Collider. We model coherent exclusive scalar and vector production at generator level in electron-gold collisions at 18 GeV by 100 GeV per nucleon. Both methods use identical weighted samples, inputs, preselection, and optimization objectives. Using only electron information, the BDT provided no consistent advantage over optimized cuts for signal selection across 11 masses for each
boson type. When both methods also use $t$, the BDT distinguishes signal from background slightly better than optimized cuts at 10 GeV for both
boson types. These results motivate further investigation of machine learning in EIC dark-
boson searches through exclusive processes where $t$ can be reconstructed.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.AI
Replication Without Persistence in Hosted LLMs: Measurement Sensitivity in Action-Time Belief Evaluation
Bhushan Kashinath Joshi
Behavioural evaluations of hosted
language models can vary because the evaluated service, the measurement instrument, or both differ across runs. We separate three validation questions: whether a prior finding recurs on fresh data under its historical configuration (replication), whether the endpoint changes when the evaluation-and-inference configuration is rebuilt under the same identifier (measurement sensitivity), and whether the finding persists across subsequently tested identifiers under one common instrument (persistence). We study these questions in Regent Chess, a sequential environment in which a hidden, mutable state is recorded exactly, allowing stated beliefs to be scored against ground truth at action time; positive endpoint values mean worse performance than a matched-uniform comparator. The previously reported Gemini 3.1 Flash-Lite deficit recurs on fresh games under its historical configuration (+0.0530, 95% CI [+0.0329,+0.0714]). In a back-to-back same-day H/R comparison under the same public identifier, the model-minus-uniform endpoint is 0.0429 lower under the rebuilt configuration (95% CI for the H-minus-R contrast [+0.0182,+0.0667]); all six configuration components vary jointly, so no component is isolated. Under rebuilt R, the prospectively frozen, interleaved same-window 4K comparison reverses sign between Gemini 3.1 and Gemini 3.7, identifiers that differ in release and product tier; additional descriptive and exploratory
cells show the same directional pattern. Any additional serving-period contribution remains unresolved (-0.0166, [-0.0483,+0.0157]). Replication, measurement sensitivity, and persistence can therefore yield different conclusions within one evaluation, motivating explicit indexing of hosted-model behavioural claims by tested identifier, serving period, measurement instrument, and inference configuration.
Read the full abstract ▾
Machine Learning
Efficient Mixture-of-Experts with Speculative Decoding via Expert Coactivation
Kumari Nishu, Han-Byul Kim, Santosh Chilkunda et al. (+13 more)
Mixture-of-Experts (MoE) models are increasingly deployed alongside Speculative Decoding (SD) to accelerate inference, but combining the two is challenging. SD improves the inference speed of dense models by verifying groups of
tokens in parallel. However, the inference speedup for SD with MoEs depends heavily on the number of
tokens being verified. Using more verification
tokens results in more experts being transferred from DRAM to the Neural Processing Unit (NPU), which increases the
memory transfer cost. This negatively impacts model runtime, as
memory transfer is typically the bottleneck in inference. In this work, we investigate the impact of MoE router design during training on the speed of MoEs with SD. We find that routers with high degrees of expert coactivation result in much faster runtimes, mitigating the impact of using more verification
tokens. Motivated by this observation, we assess the impact of various router design choices on expert coactivation and runtime using billion-parameter
transformer models. We find that combining a global load-balancing loss, shared experts, a consistency loss, and an autoregressive expert selection mechanism during training results in significantly stronger expert coactivation. This increased coactivation translates into higher overall runtime throughput: our exploration yields a model that improves throughput by 21% over MoE baselines, while maintaining on-par accuracy with the baseline MoE.
Read the full abstract ▾
All authors Kumari Nishu, Han-Byul Kim, Santosh Chilkunda, Maxwell Horton, Arnav Kundu, Mohammad Samragh, Lauren Hannah, Mohammad Sekhavat, Nikhil Bhendawade, Manuel Ciosici, Iman Mirzadeh, Keivan Alizadeh Vahid, David Harrison, Irina Belousova, Mehrdad Farajtabar, Minsik Cho
Machine Learning
Connected Content Retriever: Dense Graph Edge Features Powering Pre-Ranking at LinkedIn
Akhilesh Gupta, Sudarshan Srinivasa Ramanujam, Chirag Bhanuprasad Mehta et al. (+6 more)
In large-scale recommendation systems like the LinkedIn Feed, content generated by a member's network (connections and follows) makes up over 70% of impressions and engagement. It is therefore essential that the pre-ranking layer forwards the best possible few hundred candidates to the ranking layer. LinkedIn's professional knowledge graph carries engagement signals across both the first degree network (connections and follows) and the second-degree network: posts that a 1st-degree connection reacted to, commented on or reshared but did not author (a.k.a. stranger viral). Due to this fan out, the resulting candidate index exceeds one billion; selection of activities from the viewer's network narrows it down to roughly tens of thousands of activities that must be scored within a 120 ms p99 latency budget. We present Connected Content Retriever (CC Retriever), a pre-ranking system that scores these candidates with a full deep ranking model on GPUs at low latency. At its core is a sorted-search GPU primitive that joins dense graph affinity features (viewer to author) with document level features stored on the GPU at runtime in 5-10 ms. The shift to GPU served scoring enabled a 50x scale up of the ranking model's parameters and delivered a +2.5% lift in content time spent on the LinkedIn Feed in online experiments, significantly higher than the typical gains observed in LinkedIn Feed experiments. In this work, we describe the feature set we leverage from LinkedIn's economic graph and the model architecture used for scoring, with a particular emphasis on the online system that scales the stack.
Read the full abstract ▾
All authors Akhilesh Gupta, Sudarshan Srinivasa Ramanujam, Chirag Bhanuprasad Mehta, Reshma Asharaf Beena, Dhritiman Das, Birjodh Singh Tiwana, Bhargavkumar Kanubhai Patel, Mack Lee, Renyi Tang
Machine Learning
Complex-valued Phase-Coherent Transformers
Leona Hioki
Complex-valued
Transformers have inherited softmax
attention over the raw complex inner product. Outside natively complex domains this standard form stays near chance, and no complex
attention had been shown to correct it. We show that the match must be a scaled cosine score: L2-normalise queries and keys, so the score reads their cosine similarity and ignores their magnitudes, and hold that score at order-one scale. With this the same models train on four diagnostic tasks under two different gates; without the normalisation they stay at chance on ListOps and Needle under both gates and fall far below on the other two, and a normalised score placed at too small a scale fails as well. The resulting family of phase-coherent
Transformers (\PCT) matches or exceeds the strongest real-valued baseline across long-range
memory, positional
retrieval, hierarchical
reasoning, frequency-domain classification and physical complex signals; it shows no degradation up to depth 20; and its loss decreases log-linearly over a 61-fold range of parameters. A member of the family, complex screening combined with a phase-coherent recurrence, is the first genuinely complex-valued neural network to solve Path-X, with 91.6% of its trainable parameters complex-valued against 38.2% for S4. We record these as signs of generalisation not previously seen in complex-valued neural networks.
Read the full abstract ▾
Machine Learning
BrainWideBench: Benchmarking large-scale pretraining and across-animal transfer in multi-region neural recordings
Alexandre Andre, Shivashriganesh P. Mahato, Vinam Arora et al. (+39 more)
Advances in large-scale neural recording have made it possible to collect data across many animals and distributed
brain regions, raising the question of whether this scale can be exploited to learn general-purpose neural representations transferable across diverse downstream tasks. Yet, progress toward this goal has been limited by fragmented evaluation protocols and a narrow focus on individual task domains. Here, we present BrainWideBench, a
benchmark for evaluating across-animal transfer on multi-region neural recordings, built on the International
Brain Laboratory Brainwide Map
dataset of neural and behavioral recordings spanning 276
brain regions from 139 mice performing a sensory-guided decision-making task. The
benchmark is organized around three complementary task suites that evaluate whether learned representations support downstream decoding of behavior, can predict masked or future neural activity, and can recover biologically meaningful anatomical organization. With this
benchmark, we systematically evaluate pretraining methods across transfer settings, including finetuning on downstream objectives and zero-shot generalization to unseen animals. Our results confirm pretraining improves performance over matched single-session baselines, but we show current methods exhibit heterogeneity in transfer capabilities: gains depend strongly on the
alignment between pretraining objectives and downstream tasks. No single approach performs uniformly well across all three suites, and most methods are designed to only address a subset of them. Together, these findings suggest that learning representations that jointly generalize across behavior, dynamics, and anatomy remains an open challenge. By providing a unified and reproducible evaluation suite, BrainWideBench establishes a framework for measuring progress toward general-purpose models of the mouse
brain.
Read the full abstract ▾
All authors Alexandre Andre, Shivashriganesh P. Mahato, Vinam Arora, Keshav Balaji, Divyansha Lachi, Nanda H. Krishna, Jingyun Xiao, Yizi Zhang, Ximeng Mao, Wenrui Ma, Han Yu, International Brain Laboratory, Daniel Birman, Niccolò Bonacchi, Gaelle A. Chapuis, Joana A. Catarino, Felicia Davatolhagh, Mayo Faulkner, Laura Freitas-Silva, Fei Hu, Julia M. Huntenburg, Anup Khanal, Inês Laranjeira, Petrina Lau, Guido T. Meijer, Nathaniel J. Miska, Jean-Paul Noel, Alejandro Pan-Vazquez, Georg Raiser, Cyrille Rossant, Karolina Z. Socha, Anne E. Urai, Miles J. Wells, Steven J. West, Olivier Winter, Blake Richards, Guillaume Lajoie, Cole Hurwitz, Mehdi Azabou, Matthew R. Whiteway, Liam Paninski, Eva L. Dyer
Machine Learning · Cross-listed from cs.IR
Predictable Failure in Multi-Hop Retrieval: Score-Distributional Confidence Scoring and Abstention
Andre Bacellar
Multi-hop
retrieval failures are not uniformly distributed across queries: they cluster in structurally predictable subpopulations. We prove two results formalizing this structure. First (CWAR Reducibility): confident-failure reduction is achievable if and only if
retrieval features carry mutual information about success, a condition satisfied by LLM-judge pipelines but substantially weaker in dense-only settings, explaining the AUC-AC gap between regimes. Second (Feature Regime Complementarity): no single ANN score feature achieves best predictive performance across all failure regimes; the dominant feature differs between
datasets (query length on MuSiQue, hop-1 concentration on HoVer), and a constructive witness pair shows each is necessary in one regime and non-contributory in the other. We instantiate these principles in RegimeAbstain, which computes a
Retrieval Confidence Score (RCS), a logistic function of up to nine query-ANN structural features, all available without any additional LLM call, and uses it to implement a calibrated abstention policy. We define the Confident-Wrong-Answer Rate (CWAR) metric and evaluate across three multi-hop
benchmarks (MuSiQue, 2WikiMultiHopQA, HoVer) and two
retrieval architectures (LLM-judge and dense-only), covering five failure regimes with CWAR from 14.5% to 62.1%. RCS achieves best or co-best AUC-AC in all five conditions against eight confidence baselines. On MuSiQue (LLM-judge), RCS reduces CWAR from 39.5% to 20.6% at 50% coverage (47.8% relative reduction), with ECE=0.035. A model trained on MuSiQue transfers to 2WikiMultiHopQA with only -0.5pp AUC loss, confirming the domain-agnostic structure of regime features.
Read the full abstract ▾
Machine Learning
Benchmarking World Models for Continual Learning on Compositional Tasks
Haoyu Zhou, Joe Watson, Anson Lei et al. (+1 more)
A desirable property of a world model is the ability to learn continually across tasks, adapting to new environments without forgetting what the
agent has already learnt. In particular, the ability to retain and reuse knowledge obtained from prior experiences underpins an
agent's ability to efficiently adapt to novel environments, as the dynamics of the physical world can often be described in recurring mechanisms. However, the world model's measure of adaptation entangles two abilities: the speed and capacity to learn unseen tasks, and the reuse of knowledge already acquired, since incoming tasks carry novel content alongside what recurs. In order to isolate knowledge reuse from prior experiences, we propose a compositional continual learning
benchmark for world models in
robot manipulation. Specifically, we design each task curriculum with compositional tasks that combine aspects of the tasks seen in the sequence. We further factorise this composition along the axes of action and perception to better understand how different input modalities bottleneck knowledge reuse. We evaluate state-of-the-art world models under canonical continual learning methods, alongside a modular world model whose dynamics backbone contains explicitly reusable components. Results show that modularity balances reuse against forgetting better than conventional methods, but none solve the problem fully, leaving clear room for continual world models built to reuse without forgetting. More details are available on our project website: https://object814.github.io/Compositional-Continual-Learning/.
Read the full abstract ▾
All authors Haoyu Zhou, Joe Watson, Anson Lei, Ingmar Posner
Machine Learning
Particle Competition and Cooperation for Robust Graph Convolutional Network Learning Under Label Noise
Fabricio Breve
Graph Convolutional Networks (GCNs) are highly sensitive to label noise, since corrupted supervision can propagate through the graph and degrade learned node representations. This work proposes PCC+GCN, a hybrid framework that uses Particle Competition and Cooperation (PCC) as a graph-based label-refinement stage before GCN training. PCC identifies suspicious labeled nodes through particle domination dynamics and determines whether their labels should be preserved, removed, or reassigned before GCN training. The framework also allows the graph used by PCC to be augmented with feature-based $k$-nearest-neighbor edges, while the GCN itself is trained on the original graph structure and node features. The proposed method was evaluated on ten graph
datasets from the NoisyGL
benchmark under conventional Uniform, Pair, and Random label noise, as well as under instance-dependent label noise. A detailed hyperparameter analysis was also conducted on Cora, CiteSeer, and PubMed. Under conventional noise, PCC+GCN achieved the highest overall average accuracy and the best average rank among the evaluated methods, with an average gain of $1.67$ percentage points over the baseline GCN across the clean setting and all noisy scenarios. Under instance-dependent noise, PCC+GCN remained competitive with the best-performing robust methods while requiring substantially lower execution time, being the fastest robust method on eight of the ten
datasets. The results indicate that PCC-based label refinement provides an effective and computationally efficient preprocessing strategy for improving GCN robustness under noisy supervision.
Read the full abstract ▾
Machine Learning
Available Guardrails: Certifying Selective Prediction across ML Systems
Parivesh Priye, Yufeng Wang, Haibin Ling et al. (+1 more)
A selective predictor acts as a safety gate: it returns an output only when the prediction appears sufficiently trustworthy. Deployments increasingly require this reliability to be certified at a target precision for every reporting unit of interest, such as a tool, policy label, or patient subgroup. The main difficulty is often not whether a granted certificate is valid, but whether finite calibration data can produce one at all. As the gate becomes safer or more fine-grained, some units may receive too little evidence to certify. We make this notion of availability computable through classical exact-binomial inversion and formulate reporting-partition selection, under a fixed group order, as a dynamic program that exposes the trade-off among safety, granularity, and served traffic. The resulting frontier reveals a large population opportunity that finite-sample estimation nearly erases: a truth-informed planner gains $0.157$ mean coverage over support balancing, whereas a naive
estimator recovers only $0.005$, making recovery from finite data the central challenge. Constructing candidate partitions on one planning split and selecting among them on another recovers part of this gap, improving mean coverage over support balancing by $0.060$, with the direction reproduced in $59$ of $60$ model effects across three intent-routing
datasets and two architectures. A complementary validity-preserving lever, reallocating the familywise error budget across reporting units, recovers additional coverage both with population quantities and noisy estimates. The same frontier recurs, with predictor-specific ceilings, across LLM tool-calling, content moderation, lesion classification, and recommendation. Certified availability is therefore a plannable deployment resource that determines when a safety gate can be certified, at what granularity, and over how much traffic.
Read the full abstract ▾
All authors Parivesh Priye, Yufeng Wang, Haibin Ling, Michael Chaykowsky
Machine Learning
$\lambda$-Controlled GRPO: Turning Flow-Matching Ratio Instability into a Budgeted Resource
Yufeng Wang, Parivesh Priye, Meeshawn Marathe et al. (+1 more)
Reinforcement learning is increasingly used to align image generators with reward signals, and Flow-GRPO recently extended this paradigm to flow-matching models by treating the denoising sampler as a
stochastic policy that can be optimized from reward feedback. Training in this setting is unstable in a way specific to multi-step denoising: the policy update changes systematically across denoising steps, with importance ratios drifting below one, becoming increasingly dispersed, clipping at different rates, and leaving fewer usable samples late in training. Prior work treats these effects as separate failure modes and addresses each with a hand-tuned stabilizer. We show instead that they arise from a single per-step quantity, which we call path variance. This quantity is determined exactly by the sampler's Gaussian transition kernel and can be estimated cheaply during training. This reframes instability as a resource that can be measured and budgeted rather than a collection of symptoms to repair. Our method, $\lambda$-Controlled GRPO, calibrates importance-ratio behavior from this predicted law rather than from noisy empirical statistics, and allocates gradient effort across denoising steps according to their predicted cost. The two scales governing the update are fixed by standard policy choices rather than introduced as free tuning parameters. On a text-to-image model under two reward settings, rendering difficult target text scored by optical character recognition and matching human preferences scored by a preference model, $\lambda$-Controlled GRPO improves both text accuracy and preference reward over the strongest empirical stabilizer. It also keeps late-step path variance within its intended budget, precisely where the baseline systematically overshoots. The result is a Flow-GRPO update calibrated by its own transition law rather than stabilized after instability appears.
Read the full abstract ▾
All authors Yufeng Wang, Parivesh Priye, Meeshawn Marathe, Ramit Pahwa
Machine Learning
COMPLEX: A Closed-Form Certified Embedding of Multiparameter Persistence Modules
Sushovan Majhi, Atish Mitra, Žiga Virk et al. (+1 more)
Every multiparameter persistence vectorization we know of carries a one-sided Lipschitz upper bound and nothing below it: without a lower gauge there is no sense in which the features are faithful, and no per-prediction guarantee can be built on them. This paper supplies the missing side. COMPLEX is a closed-form, training-free
embedding of multiparameter modules – slice the module along a fixed near-diagonal net, embed each slice barcode by the certified PLACE/PALACE landmark map, concatenate. Under a checkable witnessing-slice coherence condition, holding on 100% of audited pairs on Orbit5k, a single slice carries a closed-form lower gauge: separated modules stay separated in the
embedding. With the standard upper bound this gives, to our knowledge, the first two-sided distortion bound for a multiparameter feature map, making faithfulness measurable. Measuring it, we find the floor tight within a small factor of realized distances yet operationally local: an RBF-SVM reaches 91% where 1-NN reaches 78% on the same features. Local per-prediction certification therefore fails for a structural reason common to every landmark
embedding whose lower gauge is witnessed by one coordinate. With no learned
embedding and no held-out calibration – only a cross-validated SVM head – COMPLEX sets the state of the art on both Orbit
benchmarks (91.95% on Orbit5k, 92.98% on Orbit100k), level with or above Euler-characteristic surfaces and above
transformers and graphcode. On graphs it exceeds GRIL on all four shared molecular
benchmarks with one fixed configuration, including the only multiparameter method to clear COX2's majority baseline by more than three points. Closed-form selection – of the landmark radius, the kernel (certificate-preserving), and the bifiltration set – buys further accuracy; gradient-shaped adaptation buys none.
Read the full abstract ▾
All authors Sushovan Majhi, Atish Mitra, Žiga Virk, Pramita Bagchi
Machine Learning
Abstention and Noise Filtering: Two Missing Primitives of Softmax Attention
Richard Zhe Wang
Softmax
attention has two structural gaps. A head cannot abstain, because its weights sum to one, so it outputs something even when nothing is relevant. Nor can it filter what it reads, because its output is a weighted average of value vectors, passing interference as faithfully as signal. We call these missing primitives abstention and noise filtering. Recent studies report that gating the value pathway improves pretraining but attribute the gain to different causes. We show that a value gate partly supplies both primitives, which unifies the reported causes as views of one gain. We give each primitive its own mechanism in matched models of 10M to 350M parameters and measure what each contributes. The gain from gating is almost entirely abstention at 10M, whereas by 350M filtering contributes as much as abstention, so what a study observes depends on its scale. The two benefits are largely additive, with a small overlap. A gate determined by each value alone leaves the
attention sink in place, whereas a query-controlled mechanism removes it. Injecting interference into the value reads shows that abstention and filtering protect against it in distinguishable ways. The same patterns appear in pretrained models up to 20B parameters.
Read the full abstract ▾
Machine Learning
Assessment of Machine Learning-Based Critical Heat Flux Models in the CTF Subchannel Code for Square Rod Bundle Prediction
Aidan Furlong, Vinicius de Melo Monteiro, Robert Salko et al. (+2 more)
The prediction of critical heat flux (CHF), a key safety-related quantity in
nuclear thermal hydraulics, remains an important challenge due to its direct relationship with fuel performance and reactor safety. Recent studies have demonstrated that relative to traditional empirical correlations and lookup tables (LUTs), machine learning (ML) methods can substantially improve CHF prediction accuracy. Most ML-based CHF models, however, have been developed and evaluated using tube databases, leaving their applicability to reactor-relevant rod bundle geometries largely unexplored. This study evaluates ML-based CHF models deployed within the CTF subchannel code using the Electric Power Research Institute (EPRI) rod bundle CHF database. Both pure and hybrid residual correction models are considered in local and semilocal formulations. The tube-trained ML CHF models generally transferred favorably to rod bundle applications and outperformed traditional CHF methods across most geometries and operating conditions. The local hybrid LUT model produced the strongest overall performance, and the semilocal pure ML model remained highly competitive. Comparison against the Bowring correlation, W-3 correlation, and 2006 Groeneveld LUT demonstrated that substantial improvements in rod bundle CHF prediction are possible even when models are trained exclusively on tube data. These findings provide one of the first large-scale assessments of ML-based CHF models in square rod bundles within a production-level subchannel analysis environment and support their broader application in reactor thermal hydraulic analysis.
Read the full abstract ▾
All authors Aidan Furlong, Vinicius de Melo Monteiro, Robert Salko, Juliana Pacheco Duarte, Xu Wu
Machine Learning
Time series generation with spectrally aligned latent flow matching
Camilo Carvajal Reyes and Felipe Tobar
Latent flow models have proven to be a reliable and cost-effective method for time series generation. However, the latent compression induces unwanted artefacts, such as a spectral mismatch with respect to the underlying
dataset, thus hindering their use as training surrogates. In this article, we propose a spectrally-aligned latent-flow time series generator, where the latent space for flow matching is trained to preserve dynamical properties that are relevant for the suitability of synthetic samples. We find that incorporating fine-tuning losses based on canonical signal representations such as the
Fourier, wavelet and signature transforms helps overcome these issues. The interpretability of these transformations allows us to ensure that the synthetic signals are aligned with the true ones in terms of relevant features, such as smoothness or targeted spectral content, as opposed to relying on pointwise reconstruction losses only. We compare the proposed aligned models against a base latent-flow model and the state of the art over real-world long-range univariate and multivariate
benchmark datasets. Our quantitative results validate the superiority of the proposed method in terms of its performance on metrics reflecting signal realness and computational efficiency, while being aligned to the training set with respect to its local structure.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.GT
Multiplicative Optimism for Constant Regret in Games
Ashkan Soleymani and Georgios Piliouras
We introduce Multiplicatively Optimistic Regret Matching (MORM), an uncoupled learning rule for finite general-sum games. Under simultaneous full-information self-play, every player achieves external regret $O(\sqrt n\log d)$ uniformly over all horizons, using only one-step optimism. The analysis combines a potential-based regret-matching argument with multiplicative stability and Hellinger control of strategy movement. A learning-rate safeguard additionally gives $O(\sqrt{T\log d})$ regret in the face of
adversarial utilities.
Read the full abstract ▾
Machine Learning · Cross-listed from math.ST
Schedule optimization for tau-leaping in masked discrete diffusion
Cecilia Secchi and Giacomo Zanella
Masked
diffusions are popular generative models for discrete distributions. Unlike standard autoregressive sampling, they reveal several coordinates in parallel, approximating each block's joint conditional law by a product of one-coordinate conditionals. The resulting procedure, usually called tau-leaping, reduces computational cost but introduces a factorization error ($\varepsilon_\text{fact}$), even with perfectly learned predictors. We study the resulting tradeoff between generative accuracy and computational cost, focusing on how to choose a denoising schedule to minimize $\varepsilon_\text{fact}$ for a fixed sampling budget. To do so, we establish an exact integral representation of $\varepsilon_\text{fact}$ separating the schedule from the target's dependence structure, summarized by a dependence density $\rho$. This representation yields recursive stationarity equations for optimal schedules and allows us to quantify how estimation errors in $\rho$ affect schedule selection. As the dimension $N$ and sampling budget grow, we characterize the optimal schedule and quantify the cost of random block sizes relative to a deterministic planner. We highlight a fundamental dichotomy: if $\rho$ converges uniformly to a strictly positive continuous profile as $N\to\infty$, schedule optimization can only improve the leading constant of $\varepsilon_\text{fact}$, while if $\rho$ degenerates, schedule optimization can improve the asymptotic order. Examples based on stationary processes and exchangeable mixtures illustrate these regimes.
Read the full abstract ▾
Machine Learning
RACER: Role-Aligned Competence Estimation for Human-AI Routing
Joshua Strong, Emma Sun, Alexander Capstick et al. (+3 more)
Learning to defer asks a predictive system when to act autonomously and when to defer to a human expert. Population-adaptive deferral extends this problem to unseen experts using a small context set of expert behavior. Neural context encoders such as L2D-Pop can be query-dependent, but may learn routing shortcuts tied to absolute class coordinates. Identity-Free Deferral (IFD) removes such shortcuts through role-indexed classwise competence profiles, but its estimates are constant within each class and cannot capture instance-level expert specialization. We propose RACER—Role-Aligned Competence Estimation for Routing—a role-relative framework for estimating an unseen expert's competence from context. RACER estimates the posterior-predictive probability that the expert is correct on a query under each candidate class role, then combines these estimates with the model posterior to obtain the Bayes-relevant expert-correctness probability. Nonparametric and neural kernel-pooling
estimators use candidate-role relations, shared aggregation, and symmetric summaries, excluding absolute class-identity channels. We prove coherent class-relabelling invariance, derive a Bayes-aligned deferral surrogate, and give a plug-in regret bound relating routing regret to classifier and competence-estimation error. On controlled synthetic
benchmarks, including a PathMNIST histopathology context-scaling study with simulated experts, RACER benefits from additional context under hidden subtype dependence and gives the strongest aggregate performance on a separately sampled unseen-expert split in the CIFAR-100 synthetic experiments. On the radiologist and human–AI chest-radiography
benchmarks (VinDr-CXR and CheXpert), the RACER family is competitive or best in budget-swept deferral, with calibration results varying across metrics and
datasets.
Read the full abstract ▾
All authors Joshua Strong, Emma Sun, Alexander Capstick, Pramit Saha, Cheng Ouyang, J. Alison Noble
Machine Learning
Learning to Move Cities: Deep Meta-Models and Reinforcement Policies for Calibration and Control in Urban Networks
Adewumi Augustine Adepitan, Christopher J. Haruna, Oluwasegun Adegoke et al. (+2 more)
Urban transportation networks present complex optimization challenges spanning calibration of high-fidelity simulators and real-time operational control. This paper presents a shared latent-space framework that connects simulator calibration and reinforcement learning control through a common learned representation of urban traffic dynamics. First, we develop a combinatorial MLP-autoencoder architecture that learns low-dimensional
manifolds linking simulator inputs (origin-destination demand, network parameters) to outputs (travel times, congestion patterns), enabling efficient
Bayesian optimization for calibration. This approach demonstrates superior sample efficiency compared to traditional dimension reduction methods, achieving better fit to observational data within fixed computational budgets. Second, we implement a deep Q-learning
agent with experience replay and target networks to optimize dynamic traffic assignment through scheduling and routing adjustments. In empirical evaluations on
benchmark networks, our approach reduces system-wide travel times by up to 51% compared to baseline operations. The learned latent representation is not only used to reduce the dimensionality of
Bayesian calibration, but is also incorporated into the reinforcement learning state representation, allowing the control policy to operate on compressed and calibrated traffic dynamics. This shared latent-space formulation provides a unified pathway from simulator calibration to adaptive operational control within intelligent transportation systems. Our results highlight the transformative potential of deep learning methods in urban mobility planning and management, particularly for large-scale networks where traditional optimization approaches face computational bottlenecks.
Read the full abstract ▾
All authors Adewumi Augustine Adepitan, Christopher J. Haruna, Oluwasegun Adegoke, Ayooluwatomiwa Ajiboye, Oluwatobi Oluwasakin
Machine Learning · Cross-listed from quant-ph
Guiding Agents of Quantum Games to Equilibrium using Matrix Exponential Fixed-Point Iteration
Alireza Habibi, Luis F. Abanto Leon and Setareh Maghsudi
In recent years,
quantum game theory has gained significant
attention as a framework for studying decision-making in multi-
agent systems using
quantum principles. However, computing
equilibrium strategies is challenging because the dimension of the joint
Hilbert space grows as the product of the players' local dimensions. In this paper, we consider an extended Gutoski-Watrous (EGW) game in which each player's
quantum strategy is represented by a local density
matrix. We derive tensor-contraction expressions for the payoff functions and their gradients, thereby avoiding the explicit construction of the full joint density
matrix and its computationally expensive multiplication by the payoff operators. Building on the resulting effective
Hamiltonians, we propose the
Matrix Exponential Fixed-Point Iteration with Annealing (MEFPIA)
algorithm to search for
equilibrium points in EGW games. We compare MEFPIA with the
Matrix Multiplicative Weights Update (MMWU)
algorithm in terms of convergence. For the tested instances and parameter settings, both
algorithms approach the same strategy profiles and payoffs, while MEFPIA achieves lower relative error in fewer iterations. These results indicate that MEFPIA is a promising numerical method for
equilibrium search in multi-
agent quantum games. Our findings provide important insights into the
quantum game theory's potential for addressing complex decision-making processes, as well as opening up new paths for future research and exploration in multi-
agent quantum systems.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.CR
End-to-End Hard-Label Cryptanalytic Model Extraction Using Efficient Sign Recovery
Akira Ito, Takayuki Miura and Yosuke Todo
The importance of deep neural networks (DNNs) is widely recognized, and the parameters obtained through training are regarded as valuable assets. Recently, attacks that extract these parameters using only oracle queries to a DNN have been actively studied at IACR conferences. The hard-label setting is the most challenging setting for model extraction, where an adversary can observe only the final output label, such as "dog" or "cat." At Eurocrypt 2025, Carlini et al. proposed
polynomial-time hard-label extraction of ReLU-based MLPs. However, one step of this attack process, i.e., sign recovery, requires a large number of queries and substantial computation. Implementing this step in a black-box setting remains difficult. Consequently, a fully black-box end-to-end demonstration on trained deep ReLU MLPs has remained a challenge. In this paper, we propose a new sign-recovery
algorithm based on a completely different principle from the existing method. Our method requires no dedicated queries for sign recovery. In our experiments, it achieves higher sign-recovery accuracy than the existing method. Consequently, it enables efficient sign recovery even for trained models. With our sign-recovery
algorithm, all steps of hard-label model extraction can be implemented in a black-box setting. By combining these implementations, we demonstrate end-to-end model extraction from models trained on MNIST and Fashion-MNIST, with width 16 and 4 or 6 hidden layers, achieving over 98% label agreement.
Read the full abstract ▾