Machine Learning
TracingFlow: A Simulation-Free Trajectory Inference Framework Based on Second-Order Dynamics
Yuhao Sun, Zekun Wu, Zixun Huang et al. (+1 more)
Inferring continuous system
evolution from sparse temporal snapshots is a key challenge in generative modeling and single-
cell omics. While Optimal Transport (OT) is popular, existing frameworks are largely restricted to first-order dynamics, assuming memoryless velocity fields. This limits expressiveness, as first-order systems fail to account for regulatory momentum and time-delayed responses inherent in processes like
cell differentiation. Here, we introduce TracingFlow, a simulation-free Flow Matching framework generalizing to second-order dynamics. By using neural networks to regress the acceleration field, TracingFlow provides an exact, efficient solution to the Dynamical Optimal Acceleration Transport (DOAT) problem. Unlike first-order methods yielding over-smoothed trajectories, our second-order formulation captures high-
curvature transitions and nonlinear evolutions by learning the underlying force fields. Evaluated on complex synthetic and large-scale scRNA-seq
datasets, TracingFlow achieves superior accuracy in distributional reconstruction and trajectory faithfulness. Moreover, by integrating lineage tracing priors, it recovers dynamical structures that are both mathematically optimal and biologically plausible.
Read the full abstract ▾
All authors Yuhao Sun, Zekun Wu, Zixun Huang, Peijie Zhou
Machine Learning
Designing a Robust LLM-Based Evaluation System for Agentic AI in Drug Discovery Through Human Alignment
Emma Granqvist, Rocío Mercado and Samuel Genheden
Agentic large
language model (LLM) systems are reshaping scientific workflows in chemistry and drug discovery, but evaluating their open-ended, tool-augmented outputs remains a fundamental bottleneck. The LLM-as-a-Judge paradigm has emerged as a scalable alternative, but existing drug discovery
benchmarks deploy LLM judges without validating their
alignment with human experts. In this work, we present an LLM-as-a-Judge evaluation framework for ChatInvent, an agentic drug discovery assistant deployed at AstraZeneca, with five contributions. First, we define four output-quality evaluation dimensions—Completeness, Relevancy, Structural Clarity, and Scope Adherence—alongside deterministic Tool Call Correctness checks. Second, we validate the judge through a human
alignment study with five expert annotators, comparing Gemini 3.1 Pro, Claude Opus 4.7, GPT-5, and Llama 3.1 70B as candidate judges. Third, we optimize the best-performing judge using few-shot demonstrations of human-annotated examples, improving
alignment with the human majority vote from 0.80 to 0.86. Fourth, applying the optimized judge to 70 held-out questions, we surface concrete limitations and find no strong evidence that informal phrasing degrades output quality; it may, however, still be helpful to have the LLM rewrite the original question before querying the
agent. Finally, we extend the framework to 38
adversarial questions that are ambiguous, invalid, out-of-scope or ethically sensitive, and show that the
agent's refusal behavior is guided by the stated intent of a request. Our framework provides a reusable template for human-aligned evaluation of agentic systems in scientific domains.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.CV
Chenghua Zhu, Zhaolu Kang, Qifan Shi et al. (+8 more)
Video multimodal large
language models have advanced significantly, yet fine-grained motion-temporal understanding remains fragile. The core bottleneck is not only sparse frame sampling, but also the lack of a complete temporal modeling pipeline for explicitly representing frame-to-frame change, enabling appearance-motion interaction, and optimizing temporal direction sensitivity. We propose COMET, a temporally grounded framework that systematically strengthens
video MLLMs through explicit temporal representation, appearance-motion fusion, and direction-aware optimization. Architecturally, COMET introduces a temporal motion branch built on Taylor frame differences and injects its motion evidence into the appearance stream via temporal
attention bias-enhanced cross-
attention. For optimization, COMET combines temporal prior
distillation with a forward-reverse TC-GRPO stage that turns temporal order into a direct learning signal and strengthens the model's use of directional motion patterns encoded by the temporal motion branch. The method achieves consistent overall improvements with a pronounced motion-temporal bias: on Qwen3-VL-8B, action-centric tasks (
STAR, SSv2) improve by 4.9% on average, temporal
reasoning tasks (NExT-QA, CLEVRER, LLaVA-178K) by 2.1% over BL-GRPO, while static perception tasks (PerceptionTest) remain on par. The same gain pattern also transfers to InternVL2.5-8B, indicating that COMET generalizes across model families.
Read the full abstract ▾
All authors Chenghua Zhu, Zhaolu Kang, Qifan Shi, Siyan Wu, Kehan Jiang, Lei Wei, Lianyu Hu, Guangyuan Dong, Mingbo Yang, Rui Lu, Guibo Luo
Machine Learning
RODE: A Radial-Orthogonal Decoupled Engine for Optimization
Guoxiang Xu, Bince Qu, Qi Sun et al. (+1 more)
Modern neural network training increasingly uses
matrix-aware optimizers, yet their conditioned
matrix step is typically added directly to the weight, jointly changing its norm and direction. This interaction matters because the current norm determines angular motion, while directional learning can drive norm growth and thereby alter later steps. We introduce RODE, which gives the radial and directional components separate update rules and step sizes. RODE explicitly updates the
matrix Frobenius norm through a scalar radial rule, while its directional channel performs Newton–Schulz-conditioned updates in the tangent space. Controlled GPT-2 interventions show gains from both direct norm control and RODE's directional update. Across two
language-modeling and two image-classification tasks, RODE outperforms both
Muon variants in every direct comparison and ends with lower full-model norms. At 1.5B scale, using the learning rate transferred directly from the Qwen2-style LM sweep, RODE lowers loss from 4.145 to 3.346 and final global norm from 11964 to 2183 relative to
Muon RMS, with fixed-radius RODE improving further. For Qwen3.5-9B full-parameter fine-tuning, all six optimizers use the same tuning budget and the same formal-training and evaluation settings; RODE outperforms both
Muon variants on all four evaluation tasks and attains the highest mean on GSM8K and MATH-500. Thus, decoupling radial and directional dynamics offers a more effective and controllable approach to
matrix optimization.
Read the full abstract ▾
All authors Guoxiang Xu, Bince Qu, Qi Sun, Cheng Zhuo
Machine Learning · Cross-listed from cs.IR
From a Static Multi-Level Small Semantic Codebook to a Dynamic Single-Level Large Semantic Codebook for Generative Recommendation
Tianlu Xie, Xin Ku, Mingjie Sun et al. (+8 more)
Generative recommendation represents each item with a sequence of discrete Semantic IDs (SIDs) and predicts the sequence to retrieve the next item. Typical systems use multi-level residual quantization, which increases autoregressive decoding cost and creates a large hierarchical space that may be sparsely occupied. Static codebooks also become misaligned with current traffic as new items arrive and exposure distributions change. We propose a single-level large semantic codebook that replaces multiple residual semantic codes with one semantic
token while retaining a separate collaborative disambiguation
token to reduce item collisions. We further introduce an exposure-aware dynamic update mechanism based on temporal weight decay, exponential moving-average center updates, and an exposure-weighted penalty on SID changes. We also develop an offline evaluation framework covering representation quality, code utilization, cluster load, full-SID collision, and temporal stability. On two public
datasets, the two-level SID improves mean Recall@10 by 5.0%-8.8% and mean NDCG@10 by 4.1%-5.1% for OneRec-V1, and by 7.1%-8.7% and 3.8%-8.5%, respectively, for OneRec-V2. Dynamic updating provides further gains on KuaiRec. Across three serving architectures, the shorter SID reduces estimated autoregressive-decoding FLOPs by 47.93%-48.70% and increases single-card QPS by 28.57%-47.0%. A five-day online A/B test serving 2.5% of production traffic improves the primary consumption metric by 0.792%.
Read the full abstract ▾
All authors Tianlu Xie, Xin Ku, Mingjie Sun, Yunhao Sha, Lixiang Wang, Peng Wang, Yiyu Wang, Wenjin Wu, Zhaojie Liu, Peng Jiang, Wenwu Ou
Machine Learning
Free-Probability Kernels for Zero-Rollout Hyperparameter Selection in Reservoir Computing
Sara Malacarne, Andrea Ceni and Claudio Gallicchio
Reservoir computing (RC) couples a fixed recurrent dynamical system with a trained lightweight readout, but this efficiency is partly lost during hyperparameter selection: the recurrent gain, input scale, and leakage rate determine the reservoir's stability and temporal processing regime and are usually tuned through many rollouts. We introduce a deterministic, pilot-informed selector for leaky linear reservoirs followed by coordinate-wise nonlinear features. Free probability yields cross-lag propagation coefficients that summarize how the reservoir mixes past inputs. In the large-width limit, these coefficients define a deterministic temporal kernel that approximates the finite-reservoir feature geometry. Kernel ridge regression on a short labelled pilot sequence therefore ranks candidate operating regimes without instantiating or rolling out a reservoir, and the selected configuration transfers across widths. Across ten synthetic temporal
benchmarks, zero-rollout selection obtains a mean deployment score of $0.772$, compared with $0.774$ for exhaustive simulation-based search, while avoiding $156\,600$ selection rollouts. With a small rollout budget, the proposed ranking provides the strongest mean performance at every tested budget and reaches the exhaustive reference using $4.8\%$ of its rollout cost. On four public electricity-
transformer-temperature (ETT)
forecasting datasets, five retained candidates recover the exhaustive operating point on three
datasets. On multivariate cellular-traffic
forecasting, 15 rollouts per
cell reach the 462-rollout exhaustive reference and outperform random search and
Bayesian optimization at low budgets. These results position free-probability kernels as deterministic surrogates for selecting reservoir operating regimes when validation rollouts are scarce.
Read the full abstract ▾
Machine Learning
Trojaning the Alignment: Stealthy Backdoor Attacks against Graph Foundation Models
Minhua Lin, Zhicheng Gao, Yilong Wang et al. (+3 more)
Graph Foundation Models (GFMs) on text-attributed graphs (TAGs) align graph representations with
language semantics to support transferable graph learning. Despite these advantages, the
backdoor vulnerability of GFMs on TAGs remains insufficiently understood, especially under graph-
language alignment, where graph and text representations are trained to constrain each other in a shared semantic space. Existing
backdoor attacks mainly target either the graph side or the text side, treating the two modalities independently. This makes direct adaptation ineffective: graph-only triggers can be constrained by clean text semantics, while text-only triggers alter the
language view but do not directly shift the graph representation being aligned and scored. TAGs also impose a stealth challenge because triggers are exposed as both node text and local graph structure, making incoherent trigger attributes or anomalous subgraphs easy to inspect or filter. In this paper, we propose STAG, a stealthy trojan attack framework designed for the graph-
language alignment interface of GFMs on TAGs. STAG coordinates a graph-trigger generator with a text-side soft
prompt so that trigger-attached graph representations and triggered text representations move toward the same target-class text region. To address TAG-specific stealthiness, STAG realizes trigger nodes as readable text through candidate
retrieval and regularizes the trigger-attached subgraph so that its local structure remains close to the original subgraph. Extensive experiments on multiple TAG
datasets and representative GFMs demonstrate the effectiveness and stealthiness of STAG. Our code is available at https://github.com/ventr1c/STAG.
Read the full abstract ▾
All authors Minhua Lin, Zhicheng Gao, Yilong Wang, Hanqing Lu, Xiang Zhang, Suhang Wang
Machine Learning
Jacobian-guided Noise Injection for Quantization Robustness in Large Language Models
Deepanshu Pandey, Arnav Chavan, Nahush Lele et al. (+2 more)
Quantization of Large
Language Models (LLMs) is often hindered by the sensitivity of the self-
attention mechanism to discretization errors. We identify the softmax operator as a bottleneck for quantization stability due to its sensitivity to outliers and state-dependent Jacobian. We theoretically establish that suppressing the norm of this Jacobian helps in bounding quantization-induced performance degradation. Based on this, we propose Jacobian-Guided Noise Injection, a training strategy that injects zero-mean Gaussian noise into pre-
attention logits, with variance derived directly from the Jacobian Frobenius norm. Unlike prior approaches that rely on heuristic or penalise jacobian directly, our method provides a way to identify the optimal noise variance based on the local
attention sensitivity. We evaluate the method on SOTA LLM architectures, where it demonstrates improved robustness over popular PTQ methods. Empirical analysis reveals that the proposed method gives up to +37% relative gains on Top-1 accuracy on ImageNet-1K for SigLIP and improves relative perplexity by upto 40% on WikiText for
language models in low bit quantisation settings, proving the efficacy of the approach.
Read the full abstract ▾
All authors Deepanshu Pandey, Arnav Chavan, Nahush Lele, Sankalp Dayal, Deepak Gupta
Machine Learning
Kenneth Martin, Simon Heilig, Asja Fischer et al. (+3 more)
Graph neural networks (GNNs) are routinely employed for spatiotemporal
forecasting, yet their performance across widely used
benchmark datasets is inconsistent. Here, we perform an audit of
dataset properties and baseline models to assess the quality of the
benchmarks, and the robustness of the conclusions drawn from them. Using classical statistical tools, we characterise spatiotemporal lagged dependencies in
benchmarks, and examine how temporal differencing changes these relationships and affects model rankings. Motivated by this, we re-evaluate temporal linear baselines, significantly reducing the apparent gains from GNNs on several
benchmarks, and surpassing GNNs on others. Suspecting that GNNs struggle to extract linear, node-wise signals, we find that supplying them with autoregressive residuals improves their performance particularly on non-traffic
benchmarks. Finally, controlled synthetic experiments reveal that GNNs are sensitive to heterogeneity in temporal dynamics and spatial graph interactions. Together, our findings demonstrate that baseline specification, data pre-processing and system heterogeneity shape the interpretations drawn from
benchmark rankings, informing the design and robust evaluation of GNNs.
Read the full abstract ▾
All authors Kenneth Martin, Simon Heilig, Asja Fischer, Michel F. C. Haddad, Adam M. Sykulski, Moshe Eliasof
Machine Learning · Cross-listed from eess.AS
Training DeepFilterNet with Accurate Room Acoustic Simulations Improves Single-Channel Speech Enhancement
Alessia Milo, Georg Götz, Steinar Gu{\dh}jónsson et al. (+3 more)
We investigate how the realism of synthetic room impulse response (RIR)
datasets affects the training of DeepFilterNet3 for single-channel
speech enhancement. We compare a DNS4 image-source-method (ISM) RIR
dataset with a higher-acoustic-fidelity
dataset generated using hybrid wave-based and geometrical acoustics simulation. Rather than isolating individual simulation factors, we compare complete RIR generation pipelines while keeping the enhancement model unchanged. Models are evaluated on unseen measured RIRs using objective
speech enhancement metrics and downstream automatic
speech recognition (ASR). Training with the higher-fidelity
dataset consistently yields modest improvements in objective metrics and substantially lower ASR word error rates than the ISM
dataset. Although the experiments do not attribute these gains to individual modelling components, they show that increasing the overall realism of synthetic acoustic training data improves the generalization of DeepFilterNet3 to unseen measured environments.
Read the full abstract ▾
All authors Alessia Milo, Georg Götz, Steinar Gu{\dh}jónsson, Daniel Gert Nielsen, Jesper Pedersen, Finnur Pind
Machine Learning
Training, learning and inference: unified dynamics of neural systems
Mian Wang
We define an atomic generation fact f=(u,tau,omega,z;rho), recording the origin, realized transformation, concrete occurrence, generated result and relation role. Compiled into a Generation-Fact Graph (GFG), these facts provide an AI-native, compilable scientific fact substrate preserving generation histories. We establish a GFG-based recursive scientific process in which analysis, intervention, replay and validation form facts for later cycles. Using nanoGPT, we establish unified training-learning dynamics. Training is the
evolution of a parameter-optimizer system with state and
memory: each actual training action enters the receiving state and produces a finite-amplitude nonlinear functional response conditioned by that state and target-specific update geometry. Learning is the persistent reorganization of distributed functional support by these responses; capability formation, maintenance, decline or recovery becomes observable when target-specific states are evaluated against their readout boundaries. Three primary coordinates - target-boundary state, target-specific update geometry and parameter-Adam receiving state - yield a second-order predictor operating before post-update outputs are read. On held-out runs, it achieved 91.43% accuracy and 91.49% macro-averaged recall across four transitions. We further establish inference as a frozen projection of training-learning dynamics. Component gating and rollback show causal recruitment and non-additive combination of query-conditioned support formed during training, deriving organizational conditions realized by
Attention. Controlled feedback indicates possible double-edged reinforcement effects. ResNet/CIFAR-100 and
diffusion/CIFAR-10 experiments confirm receiving-state-conditioned responses, persistent support reorganization and frozen inference projection beyond nanoGPT.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.AI
TreeWY: Speculative Verification for Gated DeltaNet Hybrids
Sneha Murthy Ghantasala
Modern open models are hybrids: most layers are linear-
attention (Gated DeltaNet, GDN) layers carrying a small fixed-size recurrent state instead of a growing key-value (KV) cache. This makes ordinary decoding
memory-efficient, but hurts speculative decoding. To verify a batch of draft
tokens and then roll back the rejected ones, today's systems snapshot the full recurrent state at every draft position for GDN layers, and those snapshots cannot be shared across branches of a draft tree, so a wide, high-acceptance tree becomes
memory-infeasible. We remove the snapshots. Using a tree-structured WY transform of the gated delta rule, we compute every draft node's output with a single triangular solve and reconstruct only the one accepted state on commit, storing a small pseudo-value
matrix instead of per-node states; the derivation depends only on the gated delta rule, not on any other architectural detail. In serving
benchmarks on two scales of one hybrid model family (Qwen3.5 35B and 397B) this cuts speculative recurrent-state
memory and KV-cache pressure at essentially identical acceptance length, turning the freed HBM into higher throughput and much lower time-to-first-
token (TTFT) wherever
memory binds, and costing a few percent where it does not. For tree width the same
memory buys affordability: a wider, higher-acceptance draft becomes possible, though not yet a throughput win.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.CL
Quantization-Aware Healing: A Practical Recipe for Recovering Compressed, 4-Bit LLMs
Bakbergen Ryskulov, Iker García-Ferrero, David Montero et al. (+5 more)
Serving large
language models cheaply increasingly means shipping models that are both structurally compressed to a fraction of their parameters and quantized to 4 bits. Together these steps degrade
reasoning, mathematics, coding, and long-context behavior enough to require a recovery, or healing, stage before deployment. The default recipe, quantization-aware training (QAT), re-fits the compressed, quantized model to hard labels; in our pipeline it converged slowly and collapsed past its peak. We adopted Quantization-Aware Healing (QAH) instead. Because a structurally compressed model is never independently trained at full precision, its bfloat16 checkpoint is a
distillation-recovered approximation of the original; QAH distills the 4-bit student directly from the original, uncompressed model. On a GPT-OSS 120B to 60B to MXFP4 pipeline, the QAH student matches or beats its bfloat16 source on 7 of 9
benchmarks at roughly 4 times less weight
memory and half the teacher's parameter count, and is released open-weight as Hypernova-60B. Against a matched QAT baseline it reaches a comparable peak about 7 times faster and stays stable under continued training, without hand-tuned early stopping. We also report deployment lessons, including a large, reproducible quality gap between distributed-training backends. Our aim is a recipe deployable without a multi-week hyper-parameter search.
Read the full abstract ▾
All authors Bakbergen Ryskulov, Iker García-Ferrero, David Montero, David Jansen, Ali Hashemi, Jezabel R. Garcia, Antonio Tiene, Román Orús
Machine Learning
Decoupling Policy Extraction for Offline Reinforcement Learning
Xuyao Lin, Yixiang Shan, Jinru Duan et al. (+7 more)
Offline RL methods commonly jointly train the actor and critic, where the critic is used to guide the actor toward higher-value actions. This coupled learning process is well motivated in online RL, where an improved actor collects new data that can further update the actor and the critic. However, training data remains fixed in offline RL, making actor-side policy improvement unable to generate new data to validate or correct the critic. Moreover, retaining this coupled paradigm leads to two related challenges. Firstly, actor updates can drift toward high-valued but potentially out-of-distribution (OOD) actions and amplify critic overestimation. Secondly, conservative value estimation or behavior-cloning regularization creates a difficult trade-off between suppressing OOD actions and selecting high-value actions within the data-supported region. Motivated by this observation, we revisit the conventional offline RL paradigm and propose decoupling policy improvement from actor training. Specifically, we train the actor solely to model the behavior distribution and perform policy improvement at inference time by reranking multiple actor-generated proposals with a separately learned critic. We refer to this paradigm as the decoupled policy extraction paradigm. Under such paradigm, the actor provides behavior-supported action candidates, while the critic performs value-based selection within this candidate set. Extensive experiments show that the decoupled policy extraction paradigm outperforms both behavior cloning and jointly learned offline RL methods, while remaining effective even with a naive Q-learning critic.
Read the full abstract ▾
All authors Xuyao Lin, Yixiang Shan, Jinru Duan, Tao Yang, Xinyu Zhao, Runyu Lei, Yiming Zhao, Jiaxin Fan, Zongbao Feng, Peng Jia
Machine Learning · Cross-listed from cs.CV
EviRank: Structured Relevance Evidence for Multimodal Image Re-ranking
Enjun Du, Siyi Liu, Zirong Chen et al. (+8 more)
Real-world image search queries are
multimodal and compositional: “find this shirt in pink” specifies an entity to retain, an attribute to modify, and context to ignore. Yet existing re-rankers either compress such multifaceted relevance into an opaque
embedding or rely on free-form chain-of-thought that easily omits or hallucinates fine-grained constraints. Drawing on rubric- and checklist-based evaluation from NLP, we recast
multimodal image re-ranking as a semantic constraint satisfaction problem and propose EviRank, which parses any query - text-only, image-only, or composed - into a unified evidence package: typed criteria across six semantic slots (e.g., entities, attributes, relations), each labelled required, forbidden, or ignorable. Re-ranking then reduces to evidence-conditioned verification, combining deterministic rubric scoring and evidence-grounded listwise comparison in a single training-free procedure. The explicit evidence can further serve as structured supervision for optionally distilling a lightweight student. Across five
benchmarks spanning text-to-image, image-to-image, and composed image
retrieval, EviRank achieves state-of-the-art performance, and the distilled student preserves over 90% of the teacher's capability at substantially lower cost.
Read the full abstract ▾
All authors Enjun Du, Siyi Liu, Zirong Chen, Xinyu Zuo, Jinwen Luo, Ruiwen Tao, Lisheng Duan, Haijin Liang, Jin Ma, Junfu Pu, Yongqi Zhang
Machine Learning
In-Cell Learning: Language Models That Update Their Own Weights in Sequence Without Changing the File They Ship
Zifeng Liu, Yaxin Lu, Xuanhan Wu et al. (+6 more)
A 4-bit quantized weight specifies a rounding
cell rather than a single full-precision value. We introduce in-
cell learning, a paradigm for writing new knowledge only within these
cells, so that re-quantizing the served weights reproduces the released integer codes and scales exactly. CellFill implements this idea with bounded trainable positions inside frozen quantization
cells and ships the update as a separate, subtractively revocable file. Across published NF4 and W4A16 releases of Qwen3 and Gemma from 1.7B to 32B parameters, CellFill writes 83-99% of a real-fact corpus while returning the stored code on every constrained weight. The injected facts generalize to paraphrases and composition, and answer 78-88% of selected PopQA questions that the released model misses. Sequential experiments show that rehearsal preserves earlier knowledge, whereas available room and new-task plasticity decline across updates. Consolidation re-quantizes the learned weights to produce a declared major version, restoring room at a measured capability cost. A six-task write-rehearse-consolidate cycle retains at least 92.8% of first learning in two 8B runs and records zero code violations over 6.9 billion constrained weights at every fold. These results define a version-management protocol in which minor updates preserve the released quantized artifact bitwise and major updates are explicit, measurable, and verifiable.
Read the full abstract ▾
All authors Zifeng Liu, Yaxin Lu, Xuanhan Wu, Zhiyong Du, Yiming Mao, Zhenhe Wang, Wenqi Shi, Zhengkun Jing, Linwei Liu
Machine Learning · Cross-listed from cs.AI
ReCurveflow: A Flow Matching Framework that Learns Curved Reaction Trajectories to Predict Transition State Geometries
Seungheun Baek, Mogan Gim and Jaewoo Kang
Predicting transition states (TS) in chemical reactions is crucial, as they provide insights into reaction mechanisms. Recent work on TS prediction have focused on flow matching supervised on straight linear paths that do not align with actual reaction trajectories. We propose a novel flow matching-based framework ReCurveflow that learns to predict TS geometries supervised on continuously curved reference paths interpolated from a full NEB-derived band of molecular geometries. We also introduce off-path correction, which grants ReCurveflow with the ability to produce corrective velocity fields when engaged off-path geometry states during inference rollout, leading to better resistance against exposure bias and accuracy in TS prediction. Across three data splits and six evaluation metrics, ReCurveflow achieves the best result on the majority of split-metric combinations against seven baselines. Qualitative analyses further show that ReCurveflow generates reaction trajectories with energy profiles that closely track the reference NEB path, provides initializations that ease the NEB optimization bottleneck, and exhibits the intended corrective behavior in its learned velocity fields. The ReCurveflow codebase is publicly available at https://github.com/dmis-lab/ReCurveflow.
Read the full abstract ▾
Machine Learning · Cross-listed from eess.SY
Sharing the Control Authority Between Deep Reinforcement Learning and Model Predictive Control: Application to Multi-Class Transportation Networks
Giray Onur, Azita Dabiri and Bart De Schutter
Transportation networks, in particular multi-class transportation networks (i.e., networks with mixed vehicle types), are complex systems that are challenging to control. Recently, Deep Reinforcement Learning (DRL), which learns control policies from interactions with the environment, and Model Predictive Control (MPC), which uses a system model to optimize control inputs, have been increasingly utilized for transportation network control. However, nonlinear system dynamics and high-dimensional state spaces in large-scale networks limit DRL's learning capacity under time-constrained training and increase MPC's computation time, hindering real-time implementation with limited computational resources. Moreover, MPC depends on an accurate network model, which is often unavailable for complex systems such as multi-class transportation networks. This paper proposes a novel DRL-MPC framework for multi-class transportation networks that divides control authority between DRL and MPC, combining DRL's fast online computation and model independence with MPC's built-in optimization and constraint-handling capabilities. In the hierarchical framework, MPC operates at the higher level and determines low-frequency control inputs whose slower update rate accommodates its high computation time, while DRL operates at the lower level and determines high-frequency control inputs using its fast online deployment. The framework is evaluated on a multi-class freeway network against a hierarchical MPC controller and a hybrid state-feedback-MPC controller, including scenarios with model mismatch and noisy traffic demands. Results show that the proposed framework outperforms the hybrid state-feedback-MPC controller, substantially reduces online computation time compared with the hierarchical MPC controller, and provides more effective constraint enforcement under model mismatch.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.CL
SAC-Copula: Quality-Preserving Watermarking for Diffusion Language Models via Smooth Correlated Gumbel Fields
Baixin Li and Haiyun He
Watermarking
diffusion language models (DLMs) requires mechanisms compatible with iterative parallel unmasking rather than autoregressive decoding. Existing sampling-based watermarking methods typically inject position-wise i.i.d. perturbations, which can be poorly aligned with DLM decoding dynamics and degrade generation quality. We propose SAC-Copula, a quality-preserving watermarking method for DLMs based on smooth, locally correlated Gumbel perturbation fields constructed via a Gaussian copula. We further develop a SAC-aware detector using covariance-aware filtering and native-sample calibration. Mechanism-level analysis shows that local correlation reduces latent perturbation roughness and better matches iterative refinement dynamics. Experiments on LLaDA show that SAC-Copula achieves a favorable quality-detectability trade-off compared with existing baselines. In particular, further evaluations on Dream-7B and additional
datasets show that SAC-Copula substantially improves PPL tail stability over the i.i.d. Gumbel baseline, while maintaining strong low-FPR detectability and competitive overall generation quality. Additional
token-edit stress tests further assess watermark robustness under controlled synchronization drift. Code is available at https://github.com/PunkyKnife/SAC-Copula.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.CY
Fine-tuning LLMs for Tourist Trajectory Prediction using Field Experiment Data
Tatsuya Amano and Hirozumi Yamaguchi
Evaluating mobility interventions at tourist destinations requires predicting visitor behavior under varying conditions. Traditional methods struggle because tourist decisions depend heavily on context like weather and fatigue, yet models cannot generalize to unobserved scenarios. Large
Language Models offer a solution by encoding commonsense knowledge about human behavior from pretraining, enabling
reasoning about context-dependent decisions, while natural
language representation flexibly integrates heterogeneous information. Fine-tuning on local trajectories adapts this general understanding to destination-specific patterns. We validate this approach using 566 trajectories from Wakayama Castle Park, Japan. Our fine-tuned Llama-3.1-8B achieves 49.1% next POI accuracy and maintains strong performance on undersampled scenarios like rainy days, demonstrating effective generalization. This establishes LLMs as high-fidelity behavior models for context-dependent tourist prediction, providing groundwork for counterfactual analysis of mobility interventions.
Read the full abstract ▾
Machine Learning
Chenghao Li, Xiao Han, Xinxin Huang et al. (+22 more)
The
matrix-aware optimizer
Muon improves large model training by balancing updates across singular directions, yet its scaling behavior and end-to-end efficiency on large
Diffusion Transformers (DiTs) remain unclear. We first establish
Muon's scaling behavior on DiTs from 1.3B to 15B parameters, showing that its optimization and generative quality advantages over AdamW persist across model scales. However, at scale, the 5-step Newton–Schulz iteration (NS5) performed at every optimization step, together with full-momentum materialization, introduces substantial computation and communication overhead that can offset
Muon's step-efficiency advantage. We introduce Periodic Row-wise
Muon, which performs a full NS5 spectral update once every \(K\) steps and applies a low compute and communication cost row-wise constrained update based on the current momentum at the remaining steps. We further co-design a distributed implementation that operates directly on sharded momentum during non-refresh steps and accelerates spectral refreshes through bucketed all-gather and communication–computation overlap. Across all scales,
Muon improves the best observed generative quality over AdamW by 12.9–19.1%. Compared with vanilla
Muon, Periodic Row-wise
Muon remains within 0.5% in best generative quality on the 1.3B–4B models and improves it by 4.5% at 9B. It reduces optimizer time by 46.9–54.3%, end-to-end step time by 15.7–24.3%, and logical communication volume by 66.7%, while reaching its respective best generative quality with 33.7–64.8% less active training time. These results show that Periodic Row-wise
Muon preserves
Muon's generative quality advantage while translating it into end-to-end training efficiency for large DiTs.
Read the full abstract ▾
All authors Chenghao Li, Xiao Han, Xinxin Huang, Wei Liu, Boyang Li, Bing Xiao, Heran Zhang, Juanma Perez Rua, Ke Xu, Kangning Liu, Linjun Kuang, Na Li, Tan Wang, Tian Xie, Wei Peng, Yang Pei, Yifan Xu, Yuanhao Zhai, Yuwei Lin, Zhe Wang, Zihao He, Daniel Li, Junbiao Tang, Ziyang Jiang, Dake Chen
Machine Learning
Resolution-Consistent Greedy Neural Approximation on Infinite-Dimensional Spaces
Pablo M. Berná, Antonio Falcó and Diego Mondéjar
We develop constructive approximation and learning guarantees for shallow neural models with infinite-dimensional inputs observed through finitely many coordinates. The analysis is based on a parameter-normalized neural dictionary and its associated weighted variation class. Within this class, the approximation error separates into a distribution-dependent coordinate-truncation term and a greedy finite-width term. For empirical regression, a fully-corrective greedy procedure yields population guarantees whose statistical complexity is uniform in the retained input resolution. The same framework extends to
Hilbert-valued responses without an explicit dependence on the output dimension. The dimension-free statements are statistical, not computational: selecting a new
neuron still requires solving a nonconvex parameter-search problem. The quasi-Polish construction underlying recent infinite-dimensional universal approximation results provides a motivating example, and synthetic experiments illustrate the predicted resolution, width, and sample-size regimes.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.AI
Neuro-Geospatial Modelling of EEG Affective States Using Literature-Informed Environmental Context
Utsav Poudel, Jagannath Aryal and Subramaniyaswamy Vairavasundaram
Environmental exposures such as air pollution and greenness have been associated with affective and cognitive outcomes, but
EEG and environmental
datasets are rarely jointly georeferenced. We investigate whether literature-informed environmental priors can serve as an auxiliary geospatial modality for
EEG-based affective-state classification when individual-level exposure data are unavailable. We combine 30-channel
EEG from the EAV
benchmark (42 participants, aged 20-30 years) with environmental representations derived from OpenAQ, Sentinel-2, Sentinel-5P, and OpenStreetMap data for Astana. A dual-tower architecture combines
EEG-Conformer representations with a graph-based environmental encoder. Because the
datasets are not co-registered, environmental context is treated as a literature-informed prior rather than measured exposure. Subject-level repeated splits, permutation and label-shuffling controls, dose-response reversal, and domain-shift experiments distinguish architecture-level gains from prior-dependent gains. The
multimodal model achieves 76.2% accuracy versus 67.4% for
EEG alone. Controls disrupting environmental-label structure retain part of this gain, indicating that the improvement is not attributable solely to environmental information. Replacing the Astana environmental distribution with an independently modeled Singapore distribution reduces accuracy to 72.8%. These findings demonstrate technical feasibility but do not establish an observed or causal exposure-affect association. The study provides a framework for future jointly collected mobile
EEG-environment studies. Implementation: https://github.com/r11up/geo-cog
Read the full abstract ▾
Machine Learning · Cross-listed from cs.GR
CubicSplat: Differentiable Vector Graphics via Error-Bounded Forward Relaxation
Chenglong Liu, Xin Zhang, Yimeng Zhu et al. (+5 more)
Vector graphics are prized for their resolution independence, compact storage, and direct editability, making differentiable optimization of their parametric primitives an attractive goal. Yet classical rasterization is discontinuous with respect to geometry, and existing remedies that smooth the forward pass demand increasingly elaborate heuristics as scene complexity grows. We trace this fragility to a gradient seesaw: design choices that improve forward
geometric exactness can systematically degrade the induced gradient signal, and vice versa. To navigate this tension we introduce CubicSplat, a differentiable vector rasterizer that replaces Bézier closest-point solvers with uniform polyline surrogates whose
geometric error is bounded at $O(S^{-2})$. The resulting static computation graph yields well-conditioned gradients by construction, while a compositing-derived visibility mechanism prunes degenerate primitives without auxiliary regularization. On DIV2K and Kodak
benchmarks CubicSplat achieves state-of-the-art reconstruction quality with over 2 dB PSNR gain in the closed-fill setting, while training up to 4x faster than prior methods. The code is available at https://github.com/CubicSplat/repo
Read the full abstract ▾
All authors Chenglong Liu, Xin Zhang, Yimeng Zhu, Liyang He, Yixiao Ma, Yu Su, Zhenya Huang, Qi Liu
Machine Learning · Cross-listed from cs.RO
Rethinking Demonstration Unlearning in Imitation Learning for Robotics
Jiazhuo Li, Yu Zhang, Yiming Fei et al. (+4 more)
Imitation learning for robotics depends on human demonstrations, some of which people may later ask to remove. Retraining without them is the natural reference, but its cost grows with policy and
dataset scale, motivating cheaper operators that edit a trained policy. Metrics inherited from machine unlearning, such as forgetting loss or a single membership attack, do not establish what an edit removed from a policy acting in closed loop. We therefore introduce a retrain-calibrated audit that reads demonstration unlearning along two axes: behavior, whether the edited policy acts like one retrained without the removed demonstrations, and evidence, whether an auditor can still detect it was trained on them. The behavior axis measures action divergence to that retrain at matched states, calibrated by a floor built from independent retrains, so a policy at the floor is as close to a retrain as retrains are to each other. The evidence axis applies a per-demonstration membership attack against a retrain null, reporting both its rank and its absolute member-loss level, since rank alone accepts operators that inflate member losses past the null. A conformal test then combines both axes into one hypothesis of joint retrain consistency, against a fleet of independent retrains large enough to reject at conventional significance. Across five preregistered conditions on three real-
robot policy classes and two simulation suites, the axes dissociate in both directions on one checkpoint, as an edit may repair task behavior while leaving evidence unchanged, or reduce evidence while moving behavior away from retraining. On the ACT arm, a redirect edit restores blind-scored
robot success to 18 of 20 trials.
Read the full abstract ▾
All authors Jiazhuo Li, Yu Zhang, Yiming Fei, Kangkang Dong, Xiaojun Zhu, Houde Liu, Jinze Tao
Machine Learning
Fuzzy-MoE: Interpretable Regime-Conditioned Expert Routing for Non-Stationary Multivariate Time Series Forecasting
Lan Guo, Jie Xiao, Zhao Su et al. (+5 more)
In non-stationary multivariate time series, different variables and samples often exhibit heterogeneous latent dynamic states, while existing deep
forecasting models usually compress them into a unified end-to-end mapping, leading to suboptimal modeling of time-varying dynamics and limited interpretability regarding which
forecasting mechanism is activated under different latent states. To overcome these limitations, we reformulate time series
forecasting as a unified framework of latent temporal state identification and interpretable expert routing, and propose Fuzzy-MoE, a fuzzy logic-based dynamic Mixture-of-Experts model. Fuzzy-MoE consists of multiple parallel expert mapping networks and a dual-view fuzzy router. By jointly exploiting local convolutional dynamics and global segmented statistics, the router infers latent temporal states and computes expert activation strengths through learnable Gaussian membership functions, enabling explicit IF-THEN rule-based expert selection. This fine-grained routing strategy allows different variables within the same sequence to activate different experts, effectively capturing heterogeneous temporal dynamics while improving model interpretability. Experimental results on multiple public time series
benchmark datasets show that Fuzzy-MoE significantly outperforms mainstream
forecasting methods in
forecasting accuracy. Moreover, fuzzy memberships and rule activations provide interpretable routing diagnostics, demonstrating the effectiveness of the proposed framework in both
forecasting performance and mechanism transparency. Unlike traditional MoE models that use black-box routing, Fuzzy-MoE`s routing is based on clear, interpretable fuzzy rules. This makes the expert selection transparent and traceable.
Read the full abstract ▾
All authors Lan Guo, Jie Xiao, Zhao Su, Jun Shen, Haoran Li, Weixia Ma, Qingguo Zhou, Binbin Yong
Machine Learning
Hidden Axis of Uncertainty: Latent-Posterior Alignment in Graph Neural Networks with Bayesian Output Layers
Suk Hoon Choi, Damdae Park, Junhyuk Choi et al. (+4 more)
Bayesian Neural Networks (BNNs) with
Bayesian output layers provide a principled and tractable framework for quantifying predictive uncertainty, yet the mechanisms shaping that uncertainty remain unclear. While conventional theory attributes uncertainty reduction to posterior contraction, the corresponding assumptions need not hold for deep models. In the Graph Neural Networks (GNNs) with
Bayesian output layers studied here, we observe that predictive uncertainty decreases as latent representations shift toward lower-variance posterior directions, even though the posterior variance does not contract. We term this behavior Latent-Posterior
Alignment (LPA) and conduct interventional experiments that support its functional role in shaping predictive uncertainty. Building on this insight, we propose
Alignment-Guided Learning (AGL), which explicitly promotes this
alignment during training. AGL effectively reduces predictive uncertainty while preserving accuracy and improves structural calibration, ensuring that the model confidence faithfully mirrors underlying data density. These findings provide a new perspective on uncertainty dynamics in GNNs with mean-field
Bayesian output layers, shifting the focus from the magnitude of the posterior to the
geometric interplay between latent and parameter spaces.
Read the full abstract ▾
All authors Suk Hoon Choi, Damdae Park, Junhyuk Choi, Hyein Jung, Changsoo Kim, Ung Lee, Kyeongsu Kim
Machine Learning · Cross-listed from cs.CL
PSK at WMT 2026 MIST: Task-Specialized QLoRA Adapters for Multilingual Summarization and Question Answering
Srikar Kashyap Pulipaka
We describe the PSK submission to the WMT 2026
Multilingual Instruction Shared Task. Our system uses the 3.35B-parameter Tiny Aya Global model with three QLoRA adapters, one for each task. The adapters are trained on
multilingual document-summary pairs, passage-based question answering, and filtered standalone question answering. The summarization data also includes scientific papers with their author-written abstracts. On our held-out split, the context and summarization adapters perform better than our multitask adapter, which was trained only on data supplied by the organizers. Results for open QA are mixed and vary with answer length and evaluation method. We therefore submit three systems with the same context and summarization adapters but different open-QA adapters.
Read the full abstract ▾
Machine Learning
Geometric Regularization for Long-Tailed Semi-Supervised Learning via Gaussian Feature Bridges
Hongyang He, Xinyuan Song, Yan Zhong et al. (+4 more)
Real-world semi-supervised learning (SSL) often encounters significant challenges with long-tailed label distributions and noisy pseudo-labels, which hinder generalization and amplify confirmation bias. In this work, we introduce a novel framework, Gaussian Bridge Consistency (GBC), to address these challenges by constructing semantic interpolation paths between unlabeled samples and high-quality class anchors. Our method maintains a dynamic Prototype Atlas that stores a diverse and evolving set of labeled and pseudo-labeled exemplars per class. For each unlabeled instance, GBC forms a class-conditional Gaussian Feature Bridge in the latent space, enabling the student model to traverse a smooth trajectory from uncertain predictions to reliable class prototypes. A bridge consistency loss is applied along this path to enforce
alignment with a geometrically interpolated target distribution. Furthermore, we propose BridgeMix, a confidence-aware feature mixing strategy that interpolates both sample and anchor pairs to amplify cross-sample generalization. Extensive experiments on CIFAR10-LT and ImageNet-LT (USB
benchmarks) validate the robustness and effectiveness of GBC under realistic long-tailed SSL settings, consistently improving long tail-class performance without sacrificing scalability.
Read the full abstract ▾
All authors Hongyang He, Xinyuan Song, Yan Zhong, Daizong Liu, Yanbin Li, Yang-fan He, Wenqiao Zhang
Machine Learning · Cross-listed from cs.AI
CDRL: Certification-Driven Reinforcement Learning for Neutrino Flavor Model Discovery
Piyush Jha, Jake Rudolph, Victoria Knapp-Pérez et al. (+3 more)
Many scientific discovery problems require searching combinatorial hypothesis spaces under complex domain constraints. Reinforcement learning (RL) offers a promising approach, but existing methods rely on scalar rewards that provide limited information about why candidate solutions fail, leading
agents to repeatedly explore invalid regions. We introduce Certification-Driven Reinforcement Learning (CDRL), a framework that leverages structured feedback from symbolic
reasoning tools. When a candidate violates domain constraints, these tools produce certificates identifying the actions responsible for failure. CDRL converts these certificates into reusable constraints that eliminate classes of invalid solutions and guide exploration toward valid regions. We evaluate CDRL on
neutrino flavor model discovery in theoretical particle physics, where the hypothesis space exceeds $10^{26}$ possible models, and compare it with the state-of-the-art RL approach previously used for this task. Across three theory spaces, CDRL achieves up to 1.95$\times$ higher valid model rates and up to 6.33$\times$ higher
neutrino model rates while evaluating up to 4$\times$ fewer candidates. We further extract 40 interpretable rules from search trajectories using a post-hoc decision-tree framework and show that reusing them as soft constraints yields gains of up to 2$\times$ in valid model rates and 3$\times$ in
neutrino model discovery across all three theory spaces. These results suggest that CDRL uncovers reusable structure in combinatorial search spaces and provides a general framework for scientific model discovery.
Read the full abstract ▾
All authors Piyush Jha, Jake Rudolph, Victoria Knapp-Pérez, Max Fieg, Aishik Ghosh, Vijay Ganesh