Machine Learning
One-Token Verification for Reasoning Correctness Estimation
Zhan Zhuang, Xiequn Wang, Zebin Chen et al. (+4 more)
Recent breakthroughs in large
language models (LLMs) have led to notable successes in complex
reasoning tasks, such as mathematical problem solving. A common strategy for improving performance is parallel thinking, in which multiple
reasoning traces are generated and the final prediction is made using aggregation schemes like majority voting or best-of-$N$ decoding. However, two key challenges persist. First, multi-sample decoding incurs substantial inference latency, especially for long-form outputs. Second, effective mechanisms for reliably assessing the correctness of individual
reasoning traces are still limited. To address these challenges, we introduce One-
Token Verification (OTV), a computational method that estimates
reasoning correctness in a single forward pass during generation. OTV is activated by a learnable
token and integrated into the LLM via low-rank adaptation to probe internal
reasoning signals through the key-value cache, supporting
token-level correctness estimation at any stage of generation without disrupting primary
reasoning. Experiments on mathematical
reasoning benchmarks demonstrate that OTV consistently surpasses existing verifiers. Additionally, OTV reduces
token usage by up to $90\%$ through correctness-guided early termination, prioritizing shorter, more reliable solutions.
Read the full abstract ▾
All authors Zhan Zhuang, Xiequn Wang, Zebin Chen, Feiyang Ye, Ying Wei, Kede Ma, Yu Zhang
Machine Learning · Cross-listed from cs.CR
BadRSSD: Backdoor Attacks on Regularized Self-Supervised Diffusion Models
Jiayao Wang, Yiping Zhang, Mohammad Maruf Hasan et al. (+5 more)
Self-supervised
diffusion models learn high-quality visual representations via latent space denoising. However, their representation layer poses a distinct threat: unlike traditional attacks targeting generative outputs, its unconstrained latent semantic space allows for stealthy
backdoors, permitting malicious control upon triggering. In this paper, we propose BadRSSD, the first
backdoor attack targeting the representation layer of self-supervised
diffusion models. Specifically, it hijacks the semantic representations of poisoned samples with triggers in Principal Component Analysis (PCA) space toward those of a target image, then controls the denoising trajectory during
diffusion by applying coordinated constraints across latent, pixel, and feature distribution spaces to steer the model toward generating the specified target. Additionally, we integrate representation dispersion regularization into the constraint framework to maintain feature space uniformity, significantly enhancing attack stealth. This approach preserves normal model functionality (high utility) while achieving precise target generation upon trigger activation (high specificity). Experiments on multiple
benchmark datasets demonstrate that BadRSSD substantially outperforms existing attacks in both FID and MSE metrics, reliably establishing
backdoors across different architectures and configurations, and effectively resisting state-of-the-art
backdoor defenses.
Read the full abstract ▾
All authors Jiayao Wang, Yiping Zhang, Mohammad Maruf Hasan, Xiaoying Lei, Jiale Zhang, Junwu Zhu, Qilin Wu, Dongfang Zhao
Machine Learning
Feature-Weighted Maximum Representative Subsampling
Tony Hauptmann and Stefan Kramer
In the social sciences, it is often necessary to debias studies and surveys before valid conclusions can be drawn. Debiasing
algorithms enable the computational removal of bias using sample weights. However, an issue arises when only a subset of features is highly biased, while the rest is already representative.
Algorithms need to strongly alter the sample distribution to manage a few highly biased features, which can in turn introduce bias into already representative variables. To address this issue, we developed a method that uses feature weights to minimize the impact of highly biased features on the computation of sample weights. Our
algorithm is based on Maximum Representative Subsampling (MRS), which debiases
datasets by aligning a non-representative sample with a representative one through iterative removal of elements to create a representative subsample. The new
algorithm, named feature-weighted MRS (FW-MRS), decreases the emphasis on highly biased features, allowing it to retain more instances for downstream tasks. The feature weights are derived from the feature importance of a domain classifier trained to differentiate between the representative and non-representative
datasets. We validated FW-MRS using eight tabular
datasets, each of which we artificially biased. Biased features can be important for downstream tasks, and focusing less on them could lead to a decline in generalization. For this reason, we assessed the generalization performance of FW-MRS on downstream tasks and found no statistically significant differences. Additionally, FW-MRS was applied to a real-world
dataset from the social sciences. The source code is available at https://github.com/kramerlab/FeatureWeightDebiasing.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.SD
AG-REPA: Causal Layer Selection for Representation Alignment in Audio Flow Matching
Pengfei Zhang, Tianxin Xie, Minghao Yang et al. (+1 more)
REPresentation
Alignment (REPA) improves the training of generative flow models by aligning intermediate hidden states with pretrained teacher features, but its effectiveness in
token-conditioned audio Flow Matching critically depends on the choice of supervised layers, which is typically made heuristically based on the depth. In this work, we introduce Attribution-Guided REPresentation
Alignment (AG-REPA), a novel causal layer selection strategy for representation
alignment in audio Flow Matching. Firstly, we find that layers that best store semantic/acoustic information (high teacher-space similarity) are not necessarily the layers that contribute most to the velocity field that drives generation, and we call it Store-Contribute Dissociation (SCD). To turn this insight into an actionable training guidance, we propose a forward-only gate ablation (FoG-A) that quantifies each layer's causal contribution via the induced change in the predicted velocity field, enabling sparse layer selection and adaptive weighting for
alignment. Across unified
speech and general-audio training (LibriSpeech + AudioSet) under different
token-conditioning
topologies, AG-REPA consistently outperforms REPA baselines. Overall, our results show that
alignment is most effective when applied to the causally dominant layers that drive the velocity field, rather than to layers that are representationally rich but functionally passive.
Read the full abstract ▾
All authors Pengfei Zhang, Tianxin Xie, Minghao Yang, Li Liu
Machine Learning
DWAFM: Dynamic Weighted Graph Structure Embedding Integrated with Attention and Frequency-Domain MLPs for Traffic Forecasting
Sen Shi, Zhichao Zhang and Yangfan He
Accurate traffic prediction is a key task for intelligent transportation systems. The core difficulty lies in accurately modeling the complex spatial-temporal dependencies in traffic data. In recent years, improvements in network architecture have failed to bring significant performance enhancements, while
embedding technology has shown great potential. However, existing
embedding methods often ignore graph structure information or rely solely on static graph structures, making it difficult to effectively capture the dynamic associations between nodes that evolve over time. To address this issue, this letter proposes a novel dynamic weighted graph structure (DWGS)
embedding method, which relies on a graph structure that can truly reflect the changes in the strength of dynamic associations between nodes over time. By first combining the DWGS
embedding with the spatial-temporal adaptive
embedding, as well as the temporal
embedding and feature
embedding, and then integrating
attention and frequency-domain multi-layer perceptrons (MLPs), we design a novel traffic prediction model, termed the DWGS
embedding integrated with
attention and frequency-domain MLPs (DWAFM). Experiments on five real-world traffic
datasets show that the DWAFM achieves better prediction performance than some state-of-the-arts.
Read the full abstract ▾
Machine Learning
Compensation-free Machine Unlearning in Text-to-Image Diffusion Models by Eliminating the Mutual Information
Xinwen Cheng, Jingyuan Zhang, Zhehao Huang et al. (+2 more)
The powerful generative capabilities of
diffusion models have raised growing
privacy and safety concerns regarding generating sensitive or undesired content. In response, machine unlearning (MU) – commonly referred to as concept erasure (CE) in
diffusion models – has been introduced to remove specific knowledge from model parameters meanwhile preserving innocent knowledge. Despite recent advancements, existing unlearning methods often suffer from excessive and indiscriminate removal, which leads to substantial degradation in the quality of innocent generations. To preserve model utility, prior works rely on compensation, i.e., re-assimilating a subset of the remaining data or explicitly constraining the divergence from the pre-trained model on remaining concepts. However, we reveal that generations beyond the compensation scope still suffer, suggesting such post-remedial compensations are inherently insufficient for preserving the general utility of large-scale generative models. Therefore, in this paper, we advocate for developing compensation-free concept erasure operations, which precisely identify and eliminate the undesired knowledge such that the impact on other generations is minimal. In technique, we propose to MiM-MU, which is to unlearn a concept by minimizing the mutual information with a delicate design for computational effectiveness and for maintaining sampling distribution for other concepts. Extensive evaluations demonstrate that our proposed method achieves effective concept removal meanwhile maintaining high-quality generations for other concepts, and remarkably, without relying on any post-remedial compensation for the first time.
Read the full abstract ▾
All authors Xinwen Cheng, Jingyuan Zhang, Zhehao Huang, Yingwen Wu, Xiaolin Huang
Machine Learning · Cross-listed from cs.AR
SoberDSE: Sample-Efficient Design Space Exploration via Learning-Based Algorithm Selection
Lei Xu, Shanshan Wang and Chenglong Xiao
High-Level Synthesis (HLS) is a pivotal electronic design automation (EDA) technology that enables the generation of hardware
circuits from high-level
language descriptions. A critical step in HLS is Design Space Exploration (DSE), which seeks to identify high-quality hardware architectures under given constraints. However, the enormous size of the design space makes DSE computationally prohibitive. Although numerous
algorithms have been proposed to accelerate DSE, our extensive experimental studies reveal that no single
algorithm consistently achieves Pareto dominance across all problem instances. Consequently, the inability of any single
algorithm to dominate all
benchmarks necessitates an automated selection mechanism to identify the best-performing DSE
algorithm for each specific case. To address this challenge, we propose the SoberDSE framework, which recommends suitable
algorithm based on
benchmark characteristics. Experimental results demonstrate that our SoberDSE framework significantly outperforms state-of-the-art heuristic-based DSE
algorithms by up to 5.7 $\times$ and state-of-the-art learning-based DSE methods by up to 4.2 $\times$. Furthermore, compared to conventional classification models, SoberDSE delivers superior accuracy in small-sample learning scenarios, with an average enhancement of 35.57%. Code and models are available at https://anonymous.4open.science/r/Sober-4377.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.AI
HiMAC: Hierarchical Macro-Micro Learning for Long-Horizon LLM Agents
Hongbo Jin, Rongpeng Zhu, Jiayu Ding et al. (+2 more)
Large
language model (LLM)
agents have recently demonstrated strong capabilities in interactive decision-making, yet they remain fundamentally limited in long-horizon tasks that require structured planning and reliable execution. Existing approaches predominantly rely on flat autoregressive policies, where high-level
reasoning and low-level actions are generated within a single
token sequence, leading to inefficient exploration and severe error propagation over extended trajectories. In this work, we propose HiMAC, a hierarchical agentic RL framework that explicitly decomposes long-horizon decision-making into macro-level planning and micro-level execution. HiMAC models
reasoning as a structured blueprint generation process followed by goal-conditioned action execution, enabling robust long-horizon planning within LLM-based
agents. To train this hierarchy efficiently, we introduce a critic-free hierarchical policy optimization paradigm that extends group-based reinforcement learning to bi-level structures through hierarchical relative advantage estimation. Furthermore, we propose an iterative co-
evolution training strategy that alternates between planner exploration and executor adaptation, mitigating the non-stationarity inherent in hierarchical learning. Extensive experiments on ALFWorld, WebShop, and Sokoban demonstrate that HiMAC consistently outperforms strong prompting and reinforcement learning baselines, achieving state-of-the-art performance and substantially improved sample efficiency across both text-based and visually grounded environments. Our results show that introducing structured hierarchy, rather than increasing model scale alone, is a key factor for enabling robust long-horizon agentic intelligence.
Read the full abstract ▾
All authors Hongbo Jin, Rongpeng Zhu, Jiayu Ding, Guibo Luo, Ge Li
Machine Learning
Forgetting is Competition: Rethinking Unlearning as Representation Interference in Diffusion Models
Ashutosh Ranjan, Vivek Srivastava, Shirish Karande et al. (+1 more)
Deployed text-to-image
diffusion models increasingly require post-hoc concept unlearning for copyright claims, artist opt-outs, safety updates, and protected-content mitigation without full retraining. A central challenge is erase-retain imbalance, aggressive updates suppress targets but damage shared capabilities, while conservative or anchor-based updates preserve quality yet leave concepts recoverable through related, compositional, paraphrased, or
adversarial prompts. Inspired by retroactive interference, we propose SurgUn, which treats forgetting as controlled competition rather than direct deletion or one-to-one reassignment. SurgUn instantiates retroactive concept interference via distractor-conditioned gradient competition: target-gradient ascent weakens target-conditioned denoising or flow-matching behavior, while descent over a semantically diverse distractor set introduces competing non-target trajectories under the same
prompt context. This redistributes outputs across multiple non-target modes instead of collapsing to a single proxy. To limit collateral forgetting through shared pathways, SurgUn adds pixel-grounded weight-space localization, a lightweight diagnostic that selects
attention blocks by generated-image erase-retain behavior, exploiting the asymmetry that suppression is broadly achievable whereas retention is block-selective. Across UnlearnCanvas, IP-character erasure, Holistic Unlearning, EraseBench, and Ring-A-Bell on Stable
Diffusion v1.5, SDXL, and SANA-1.5, SurgUn achieves a stronger erase-retain balance than baselines. Ablations show that diverse distractors, contrastive competition, and localization are all necessary for robust suppression while preserving related and unrelated concepts.
Read the full abstract ▾
All authors Ashutosh Ranjan, Vivek Srivastava, Shirish Karande, Murari Mandal
Machine Learning
Intent-Context Synergy Reinforcement Learning for Autonomous UAV Decision-Making in Air Combat
Jiahao Fu and Feng Yang
Autonomous
UAV infiltration in dynamic contested environments remains a significant challenge due to the partially observable nature of threats and the conflicting objectives of mission efficiency versus survivability. Traditional Reinforcement Learning (RL) approaches often suffer from myopic decision-making and struggle to balance these trade-offs in real-time. To address these limitations, this paper proposes an Intent-Context Synergy Reinforcement Learning (ICS-RL) framework. The framework introduces two core innovations: (1) An LSTM-based Intent Prediction Module that forecasts the future trajectories of hostile units, transforming the decision paradigm from reactive avoidance to proactive planning via state augmentation; (2) A Context-Analysis Synergy Mechanism that decomposes the mission into hierarchical sub-tasks (safe cruise, stealth planning, and hostile breakthrough). We design a heterogeneous ensemble of Dueling DQN
agents, each specialized in a specific tactical context. A dynamic switching controller based on Max-Advantage values seamlessly integrates these
agents, allowing the
UAV to adaptively select the optimal policy without hard-coded rules. Extensive simulations demonstrate that ICS-RL significantly outperforms baselines (Standard DDQN) and traditional methods (PSO, Game Theory). The proposed method achieves a mission success rate of 88% and reduces the average exposure frequency to 0.24 per episode, validating its superiority in ensuring robust and stealthy penetration in high-dynamic scenarios.
Read the full abstract ▾
Machine Learning · Cross-listed from stat.ML
Random Features for Operator-Valued Kernels: Bridging Kernel Methods and Neural Operators
Mike Nguyen and Nicole Mücke
In this work, we investigate the generalization properties of random feature methods. Our analysis extends prior results for Tikhonov regularization to a broad class of spectral regularization techniques and further generalizes the setting to operator-valued kernels. This unified framework enables a rigorous theoretical analysis of neural operators and neural networks through the lens of the Neural Tangent Kernel (NTK). In particular, it allows us to establish optimal learning rates and provides a good understanding of how many
neurons are required to achieve a given accuracy. Furthermore, we establish minimax rates in the well-specified case and also in the misspecified case, where the target is not contained in the reproducing kernel
Hilbert space. These results sharpen and complete earlier findings for specific kernel
algorithms.
Read the full abstract ▾
Machine Learning · Cross-listed from stat.ML
Learning with the Nash-Sutcliffe loss
Hristos Tyralis and Georgia Papacharalampous
The Nash-Sutcliffe efficiency ($\text{NSE}$) is a widely used, positively oriented relative measure for evaluating forecasts across multiple time series. However, it lacks a decision-theoretic foundation for this purpose. To address this, we examine its negatively oriented counterpart, which we refer to as Nash-Sutcliffe loss, defined as $L_{\text{NS}} = 1 - \text{NSE}$. We prove that $L_{\text{NS}}$ is strictly consistent for an elicitable and identifiable multi-dimensional functional, which we name the Nash-Sutcliffe functional. This functional is a data-weighted component-wise mean. The common practice of maximizing the average $\text{NSE}$ across multiple series is the sample analog of minimizing the expected $L_{\text{NS}}$. Consequently, this operation implicitly assumes that all series originate from a single non-stationary,
stochastic process. We introduce Nash-Sutcliffe linear regression, a multi-dimensional model estimated by minimizing the average $L_{\text{NS}}$, which reduces to a data-weighted least squares formulation. By reorienting the sample average loss function, we extend the previously proposed evaluation and estimation framework to
forecasting multiple stationary dependent time series with differing
stochastic properties. This constitutes a more natural empirical implementation of the $\text{NSE}$ than the earlier formulation. Our results establish a decision-theoretic foundation for $\text{NSE}$-based model estimation and forecast evaluation in large
datasets, while further clarifying the benefits of global over local machine learning models.
Read the full abstract ▾
Machine Learning
Stabilizing Policy Optimization via Logits Convexity
Hongzhan Chen, Tao Yang, Yuhua Zhu et al. (+3 more)
While reinforcement learning (RL) has been central to the recent success of large
language models (LLMs), RL optimization is notoriously unstable, especially when compared to supervised fine-tuning (SFT). In this work, we investigate the stability gap between SFT and RL from a gradient-based perspective, and show that the convexity of the SFT loss with respect to model logits plays a key role in enabling stable training. Our theoretical analysis demonstrates that this property induces favorable gradient directionality during optimization. In contrast, Proximal Policy Optimization (PPO), a widely adopted policy gradient
algorithm utilizing a clipped surrogate objective, lacks this stabilizing property. Motivated by this observation, we propose Logits Convex Optimization (LCO), a simple yet effective policy optimization framework that aligns the learned policy with an optimal target derived from the original RL objective, thereby emulating the stabilizing effects of logits-level convexity. Extensive experiments across multiple model families show that our LCO framework consistently improves training stability and outperforms conventional RL methods on a broad range of
benchmarks.
Read the full abstract ▾
All authors Hongzhan Chen, Tao Yang, Yuhua Zhu, Shiping Gao, Xiaojun Quan, Ting Yao
Machine Learning
When Does Margin Clamping Affect Training Variance? Dataset-Dependent Effects in Contrastive Forward-Forward Learning
Joshua Steier
Contrastive Forward-Forward (CFF) learning trains
Vision Transformers layer by layer against supervised contrastive objectives. CFF training can be sensitive to random seed, but the sources of this instability are poorly understood. We focus on one implementation detail: the positive-pair margin in the contrastive loss is applied through saturating similarity clamping, $\min(s + m,\, 1)$. We prove that an alternative formulation, subtracting the margin after the log-probability, is gradient-neutral under the mean-over-positives reduction. On CIFAR-10 ($2 \times 2$ factorial, $n{=}7$ seeds per
cell), clamping produces $5.90\times$ higher pooled test-accuracy variance ($p{=}0.003$) with no difference in mean accuracy. Analyses of clamp activation rates, layerwise gradient norms, and a reduced-margin probe point to saturation-driven gradient truncation at early layers. The effect does not transfer cleanly to other
datasets: on CIFAR-100, SVHN, and Fashion-MNIST, clamping produces equal or lower variance. Two factors account for the discrepancy. First, positive-pair density per batch controls how often saturation occurs. Second, task difficulty compresses seed-to-seed spread when accuracy is high. An SVHN difficulty sweep confirms the interaction on a single
dataset, with the variance ratio moving from $0.25\times$ at high accuracy to $16.73\times$ under aggressive augmentation. In moderate-accuracy regimes with many same-class pairs per batch, switching to the gradient-neutral subtraction reference removes this variance inflation at no cost to mean accuracy. Measuring the layer-0 clamp activation rate serves as a simple check for whether the problem applies.
Read the full abstract ▾
Machine Learning · Cross-listed from math.OC
Non-Rectangular Average-Reward Robust MDPs: Optimal Policies and Their Transient Values
Shengbo Wang and Nian Si
We study non-rectangular robust Markov decision processes under the average-reward criterion, where the ambiguity set couples transition probabilities across states and the adversary commits to a stationary kernel for the entire horizon. We show that any history-dependent policy achieving sublinear expected regret uniformly over the ambiguity set is robust-optimal, and that the robust value admits a minimax representation as the infimum over the ambiguity set of the classical optimal gains, without requiring any form of rectangularity or robust dynamic programming principle. Under the weak communication assumption, we establish the existence of such policies by converting high-probability regret bounds from the average-reward reinforcement learning literature into the expected-regret criterion. We then introduce a transient-value framework to evaluate finite-time performance of robust optimal policies, proving that average-reward optimality alone can mask arbitrarily poor transients and deriving regret-based lower bounds on transient values. Finally, we construct an epoch-based policy that combines an optimal stationary policy for the worst-case model with an anytime-valid sequential test and an online learning fallback, achieving a constant-order transient value.
Read the full abstract ▾
Machine Learning · Cross-listed from stat.ML
Time-Aware Latent Space Bayesian Optimization
Tuan A. Vu, Julien Martinelli and Harri Lähdesmäki
Latent-space
Bayesian optimization (LSBO) extends
Bayesian optimization to structured domains, such as molecular design, by searching in the continuous latent space of a generative model. However, most LSBO methods assume a fixed objective, whereas real design campaigns often face temporal drift (e.g., evolving preferences or shifting targets). Bringing time-varying BO into LSBO is nontrivial: drift can affect not only the surrogate, but also the latent search space geometry induced by the representation. We propose Time-Aware Latent-space
Bayesian Optimization (TALBO), which incorporates time in both the surrogate and the learned generative representation via a GP-prior variational autoencoder, yielding a latent space aligned as objectives evolve. To evaluate timevarying LSBO systematically, we adapt widely used molecular design tasks to drifting multi-property objectives and introduce metrics tailored to changing targets. Across these
benchmarks, TALBO consistently outperforms strong LSBO baselines and remains robust across drift speeds and design choices, while remaining competitive under actually time-invariant objectives.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.IT
Curvature-Weighted Capacity Allocation: A Minimum Description Length Framework for Layer-Adaptive Large Language Model Optimization
Theophilus Amaefuna, Hitesh Vaidya, Anshuman Chhabra et al. (+1 more)
Layer-wise capacity in large
language models is highly non-uniform: some layers contribute disproportionately to loss reduction, whereas others are nearly redundant. Existing layer-scoring methods provide sensitivity estimates but do not give a principled rule for converting those estimates into allocation or
pruning decisions under a global hardware budget. We introduce a
curvature-aware, MDL-inspired framework built around the layer gain $\zeta_k^2=g_k^\top\widetilde H_{kk}^{-1}g_k$. This quantity equals twice the maximal decrease predicted by the regularized layer-restricted quadratic model and incorporates inverse local
curvature; it is therefore a local surrogate for reducible risk, not a universal dominance claim over gradient-norm scores. After normalizing the gains into scores $q_k$, we formulate two convex programs: one allocates expert slots under diminishing returns, and the other assigns layer-wise
pruning ratios while protecting high-score layers. Both continuous programs have unique globally optimal solutions characterized by one dual variable and computable in $O(K\log(1/\varepsilon))$ time by bisection. We also prove a quadratic transfer-regret bound: when source and target score vectors differ by at most $\delta$, the target surrogate cost of the transferred decision is within $O(\delta^2)$ of the target optimum. Experiments on Mistral-7B and Gemma-7B show clear allocation gains in some settings and competitive, though mixed,
pruning performance. The framework therefore replaces an empirical score-to-decision heuristic with a budget-feasible optimization procedure whose guarantees apply to the stated continuous surrogates. Code is available on github repo - [TKAI-LAB-Mali/
Curvature-Weighted-Capacity-Allocation](https://github.com/TKAI-LAB-Mali/
Curvature-Weighted-Capacity-Allocation.git)
Read the full abstract ▾
All authors Theophilus Amaefuna, Hitesh Vaidya, Anshuman Chhabra, Ankur Mali
Machine Learning
Principled Fast and Meta Knowledge Learners for Continual Reinforcement Learning
Ke Sun, Hongming Zhang, Jun Jin et al. (+4 more)
Inspired by the human learning and
memory system, particularly the interplay between the hippocampus and cerebral cortex, this study proposes a dual-learner framework comprising a fast learner and a meta learner to address continual Reinforcement Learning~(RL) problems. These two learners are coupled to perform distinct yet complementary roles: the fast learner focuses on knowledge transfer, while the meta learner ensures knowledge integration. In contrast to traditional multi-task RL approaches that share knowledge through average return maximization, our meta learner incrementally integrates new experiences by explicitly minimizing catastrophic forgetting, thereby supporting efficient cumulative knowledge transfer for the fast learner. To facilitate rapid adaptation in new environments, we introduce an adaptive meta warm-up mechanism that selectively harnesses past knowledge. We conduct experiments in various pixel-based and continuous control
benchmarks, revealing the superior performance of continual learning for our proposed dual-learner approach relative to baseline methods. The code is released in https://github.com/datake/FAME.
Read the full abstract ▾
All authors Ke Sun, Hongming Zhang, Jun Jin, Chao Gao, Xi Chen, Wulong Liu, Linglong Kong
Machine Learning
Evaluating AI Grading on Real-World Handwritten College Mathematics: A Large-Scale Study Toward a Benchmark
Zhiqi Yu, Xingping Liu, Haobin Mao et al. (+4 more)
Grading in large undergraduate STEM courses often yields minimal feedback due to heavy instructional workloads. We present a large-scale empirical study of AI grading on real, handwritten single-variable calculus work from UC Irvine. Using OCR-conditioned large
language models with structured, rubric-guided prompting, our system produces scores and formative feedback for thousands of free-response quiz submissions from nearly 800 students. In a setting with no single ground-truth label, we evaluate performance against official teaching-assistant grades, student surveys, and independent human review, finding strong
alignment with TA scoring and a large majority of AI-generated feedback rated as correct or acceptable across quizzes. Beyond calculus, this setting highlights core challenges in OCR-conditioned mathematical
reasoning and partial-credit assessment. We analyze key failure modes, propose practical rubric- and
prompt-design principles, and introduce a multi-perspective evaluation protocol for reliable, real-course deployment. Building on the
dataset and evaluation framework developed here, we outline a standardized
benchmark for AI grading of handwritten mathematics to support reproducible comparison and future research.
Read the full abstract ▾
All authors Zhiqi Yu, Xingping Liu, Haobin Mao, Mingshuo Liu, Long Chen, Jack Xin, Yifeng Yu
Machine Learning
Probabilistic Learning and Generation in Deep Sequence Models
Wenlong Chen
Despite exceptional predictive performance of Deep sequence models (DSMs), the main concern of their deployment centers around the lack of uncertainty awareness. In contrast, probabilistic models quantify the uncertainty associated with unobserved variables with rules of probability. Notably,
Bayesian methods leverage Bayes' rule to express our belief of unobserved variables in a principled way. Since exact
Bayesian inference is computationally infeasible at scale, approximate inference is required in practice. Two major bottlenecks of
Bayesian methods, especially when applied in deep neural networks, are prior specification and approximation quality. In Chapter 3 & 4, we investigate how the architectures of DSMs themselves can be informative for the design of priors or approximations in probabilistic models. We first develop an approximate
Bayesian inference method tailored to the
Transformer based on the similarity between
attention and sparse Gaussian process. Next, we exploit the long-range
memory preservation capability of HiPPOs (High-order
Polynomial Projection Operators) to construct an interdomain inducing point for Gaussian process, which successfully memorizes the history in online learning. In addition to the progress of DSMs in predictive tasks, sequential generative models consisting of a sequence of latent variables are popularized in the domain of deep generative models. Inspired by the explicit self-supervised signals for these latent variables in
diffusion models, in Chapter 5, we explore the possibility of improving other generative models with self-supervision for their sequential latent states, and investigate desired probabilistic structures over them. Overall, this thesis leverages inductive biases in DSMs to design probabilistic inference or structure, which bridges the gap between DSMs and probabilistic models, leading to mutually reinforced improvement.
Read the full abstract ▾
Machine Learning
Knowledge without Wisdom: Measuring Misalignment between LLMs and Intended Impact
Michael Hardy and Yunsung Kim
LLMs increasingly excel on AI
benchmarks, but doing so does not guarantee validity for downstream tasks. This study contrasts LLM
alignment on
benchmarks, downstream tasks, and, importantly the intended impact of those tasks. We evaluate the performance of leading LLMs (i.e., generative pre-trained base models) on difficult-to-verify tasks of the teaching and learning of schoolchildren. Across all LLMs, inter-model behaviors on disparate tasks correlate higher than they do with expert human behaviors on target tasks. These biases shared across LLMs are poorly aligned with downstream measures of teaching quality and often negatively aligned with the intended impact of student learning outcomes. Further, we find multi-model ensembles, both unanimous model voting and expert-weighting by
benchmark performance, further exacerbate misalignment with learning. We measure that selection of LLM and/or prompting strategy only reliably accounts for $15\%$ of all measured misalignment error and that variation in misalignment error is shared across LLMs, suggesting that common pretraining accounts for much of the misalignment in these tasks. We demonstrate methods for robustly measuring
alignment of complex tasks and provide unique insights into practical applications of LLMs in high-noise contexts.
Read the full abstract ▾
Machine Learning
Active Flow Matching
Yashvir S. Grewal, Daniel M. Steinberg, Thang D. Bui et al. (+2 more)
Discrete
diffusion and flow matching models capture complex, non-additive and non-autoregressive structure in high-dimensional objective landscapes through parallel, iterative refinement. However, their implicit generative nature precludes direct integration with principled variational frameworks for online black-box optimisation, such as variational search distributions (VSD) and conditioning by adaptive sampling (CbAS). We introduce Active Flow Matching (AFM), which reformulates variational objectives to operate on conditional endpoint distributions along the flow, enabling gradient-based steering of flow models toward high-fitness regions while preserving the rigour of VSD and CbAS. We derive forward and reverse Kullback-Leibler (KL) variants using self-normalised importance sampling. Across a suite of online
protein and small molecule design tasks, forward-KL AFM consistently performs competitively compared to state-of-the-art baselines, demonstrating effective exploration-exploitation under tight experimental budgets.
Read the full abstract ▾
All authors Yashvir S. Grewal, Daniel M. Steinberg, Thang D. Bui, Cheng Soon Ong, Edwin V. Bonilla
Machine Learning
MultiPUFFIN: A Multimodal Domain-Constrained Foundation Model for Molecular Property Prediction of Small Molecules
Idelfonso B. R. Nogueira, Carine M. Rebello, Mumin Enis Leblebici et al. (+1 more)
MultiPUFFIN is a domain-informed
multimodal foundation model for predicting thermophysical properties of small molecules, addressing a critical gap in chemical engineering, drug discovery, and materials science. Existing molecular foundation models pretrain on millions of molecules to learn general-purpose representations, but their standard MLP output layers impose no physical constraints, vapor pressure predictions may violate monotonic temperature dependence, and viscosity curves may lack the functional form required by process simulators. Domain-informed approaches that guarantee thermodynamic consistency have remained limited to single properties and small
datasets, whereas
multimodal foundation models have focused on biological activity rather than thermophysical properties. MultiPUFFIN fills this gap by fusing SMILES sequences, 2D molecular graphs, and 3D conformer geometries through bidirectional cross-modal
attention and gated fusion, supplemented by auxiliary encoders for experimental conditions and molecular descriptors. The backbone is pretrained on 500,000 unlabelled PubChem molecules using three complementary self-supervised objectives. A condition-aware refinement stack of five conditioners (temperature, pH, pressure, polymorph, and measurement method) routes each property to a four-head tournament that selects the best-performing thermodynamically informed head for that property. MultiPUFFIN achieves a mean test R2 of 0.784 and outperforms fine-tuned ChemBERTa-2 on all nine properties despite training on roughly 2,000x fewer labeled molecules.
Read the full abstract ▾
All authors Idelfonso B. R. Nogueira, Carine M. Rebello, Mumin Enis Leblebici, Erick Giovani Sperandio Nascimento
Machine Learning
Navigating Time's Possibilities: Plausible Counterfactual Explanations for Multivariate Time-Series Forecast through Genetic Algorithms
Gianlucca Zuin and Adriano Veloso
Counterfactual learning has become promising for understanding and modeling causality in complex and dynamic systems. This paper presents a novel method for counterfactual learning in the context of multivariate time series analysis and forecast. The primary objective is to uncover hidden causal relationships and identify potential interventions to achieve desired outcomes. The proposed methodology integrates genetic
algorithms and rigorous causality tests to infer and validate counterfactual dependencies within temporal sequences. More specifically, we employ Granger causality to enhance the reliability of identified causal relationships, rigorously assessing their statistical significance. Then, genetic
algorithms, in conjunction with
quantile regression, are used to exploit these intricate causal relationships to project future scenarios. The synergy between genetic
algorithms and causality tests ensures a thorough exploration of the temporal dynamics present in the data, revealing hidden dependencies and enabling the projection of outcomes under hypothetical interventions. We evaluate the performance of our
algorithm on real-world data, showcasing its ability to handle complex causal relationships, revealing meaningful counterfactual insights, and allowing for the prediction of outcomes under hypothetical interventions.
Read the full abstract ▾
Machine Learning
GeMi: A Graph-based, Multimodal Recommendation System for Narrative Scroll Paintings
Haimonti Dutta, Pruthvi Moluguri, Jin Dai et al. (+1 more)
Recommendation Systems are effective in managing the ever-increasing amount of
multimodal data available today and help users discover interesting new items. These systems can handle various media types such as images, text, audio, and
video data, and this has made it possible to handle content-based recommendation utilizing features extracted from items while also incorporating user preferences. Graph Neural Network (GNN)-based recommendation systems are a special class of recommendation systems that can handle relationships between items and users, making them particularly attractive for content-based recommendations. Their popularity also stems from the fact that they use advanced machine learning techniques, such as deep learning on graph-structured data, to exploit user-to-item interactions. The nodes in the graph can access higher-order neighbor information along with state-of-the-art
vision-
language models for processing
multimodal content, and there are well-designed
algorithms for
embedding, message passing, and propagation. In this work, we present the design of a GNN-based recommendation system on a novel data set collected from field research. Designed for an endangered performing art form, the recommendation system uses
multimodal content (text and image data) to suggest similar paintings for viewing and purchase. To the best of our knowledge, there is no recommendation system designed for narrative scroll paintings – our work therefore serves several purposes, including art conservation, a data storage system for endangered art objects, and a state-of-the-art recommendation system that leverages both the novel characteristics of the data and preferences of the user population interested in narrative scroll paintings.
Read the full abstract ▾
All authors Haimonti Dutta, Pruthvi Moluguri, Jin Dai, Saurabh Amarnath Mahindre
Machine Learning · Cross-listed from cs.IR
Tiny-Critic RAG: Empowering Agentic Fallback with Parameter-Efficient Small Language Models
Yichao Wu, Penghao Liang, Yafei Xiang et al. (+5 more)
Retrieval-Augmented Generation (RAG) grounds Large
Language Models (LLMs) to mitigate factual
hallucinations. Recent paradigms shift from static pipelines to Modular and Agentic RAG frameworks, granting models autonomy for multi-hop
reasoning or self-correction. However, current reflective RAG heavily relies on massive LLMs as universal evaluators. In high-throughput systems, executing complete forward passes for billion-parameter models merely for binary routing introduces severe computational redundancy. Furthermore, in autonomous
agent scenarios, inaccurate
retrieval causes models to expend excessive
tokens on spurious
reasoning and redundant tool calls, inflating Time-to-First-
Token (TTFT) and costs. We propose Tiny-Critic RAG, decoupling evaluation by deploying a parameter-efficient Small
Language Model (SLM) via Low-Rank Adaptation (LoRA). Acting as a deterministic gatekeeper, Tiny-Critic employs constrained decoding and non-thinking inference modes for ultra-low latency binary routing. Evaluations on noise-injected
datasets demonstrate Tiny-Critic RAG achieves routing accuracy comparable to GPT-4o-mini while reducing latency by an order of magnitude, establishing a highly cost-effective paradigm for
agent deployment.
Read the full abstract ▾
All authors Yichao Wu, Penghao Liang, Yafei Xiang, Mengwei Yuan, Jianan Liu, Jing Yang, Xianyou Li, Weiran Yan