Peer Revenue Past citations are no guarantee of future returns

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

Computer Science Machine Learning

cs.LG Watch Machine Learning (sign in to use)

Friday 20 March

Machine Learning

Two-Time-Scale Learning Dynamics: A Population View of Neural Network Training

Population-based learning paradigms, including evolutionary strategies, Population-Based Training (PBT), and recent model-merging methods, combine fast within-model optimisation with slower population-level adaptation. Despite their empirical success, a general mathematical description of the resulting collective training dynamics remains incomplete. We introduce a theoretical framework for neural network training based on two-time-scale population dynamics. We model a population of neural networks as an interacting agent system in which network parameters evolve through fast noisy gradient updates of SGD/Langevin type, while hyperparameters evolve through slower selection–mutation dynamics. We prove the large-population limit for the joint distribution of parameters and hyperparameters and, under strong time-scale separation, derive a selection–mutation equation for the hyperparameter density. For each fixed hyperparameter, the fast parameter dynamics relaxes to a Boltzmann–Gibbs measure, inducing an effective fitness for the slow evolution. The averaged dynamics connects population-based learning with bilevel optimisation and classical replicator–mutator models, yields conditions under which the population mean moves toward the fittest hyperparameter, and clarifies the role of noise and diversity in balancing optimisation and exploration. Numerical experiments illustrate both the large-population regime and the reduced two-time-scale dynamics, and indicate that access to the effective fitness, either in closed form or through population-level estimation, can improve population-level updates.
Read the full abstract

Machine Learning

Quantifying Gate Contribution in Quantum Feature Maps for Scalable Circuit Optimization

Quantum machine learning offers promising advantages for classification tasks, but noise, decoherence, and connectivity constraints in current devices continue to limit the efficient execution of feature map-based circuits. Gate Assessment and Threshold Evaluation (GATE) is presented as a circuit optimization methodology that reduces quantum feature maps using a novel gate significance index. This index quantifies the relevance of each gate by combining fidelity, entanglement, and sensitivity. It is formulated for both simulator/emulator environments, where quantum states are accessible, and for real hardware, where these quantities are estimated from measurement results and auxiliary circuits. The approach iteratively scans a threshold range, eliminates low-contribution gates, generates optimized quantum machine learning models, and ranks them based on accuracy, runtime, and a balanced performance criterion before final testing. The methodology is evaluated on real-world classification datasets using two representative quantum machine learning models, PegasosQSVM and Quantum Neural Network, in three execution scenarios: noise-free simulation, noisy emulation derived from an IBM backend, and real IBM quantum hardware. The structural impact of gate removal in feature maps is examined, compatibility with noise-mitigation techniques is studied, and the scalability of index computation is evaluated using approaches based on density matrices, matrix product states, tensor networks, and real-world devices. The results show consistent reductions in circuit size and runtime and, in many cases, preserved or improved predictive accuracy, with the best trade-offs typically occurring at intermediate thresholds rather than in the baseline circuits or in those compressed more aggressively.
Read the full abstract

All authors F. Rodríguez-Díaz, D. Gutiérrez-Avilés, A. Troncoso, F. Martínez-Álvarez

Machine Learning

Scalable Learning of Multivariate Distributions via Coresets

Efficient and scalable non-parametric or semi-parametric regression analysis and density estimation are of crucial importance to the fields of statistics and machine learning. However, available methods are limited in their ability to handle large-scale data. We address this issue by developing a novel coreset construction for multivariate conditional transformation models (MCTMs) to enhance their scalability and training efficiency. To the best of our knowledge, these are the first coresets for semi-parametric distributional models. Our approach yields substantial data reduction via importance sampling. It ensures with high probability that the log-likelihood remains within multiplicative error bounds of $(1\pm\varepsilon)$ and thereby maintains statistical model accuracy. Compared to conventional full-parametric models, where coresets have been incorporated before, our semi-parametric approach exhibits enhanced adaptability, particularly in scenarios where complex distributions and non-linear relationships are present, but not fully understood. To address numerical problems associated with normalizing logarithmic terms, we follow a geometric approximation based on the convex hull of input data. This ensures feasible, stable, and accurate inference in scenarios involving large amounts of data. Numerical experiments demonstrate substantially improved computational efficiency when handling large and complex datasets, thus laying the foundation for a broad range of applications within the statistics and machine learning communities.
Read the full abstract

All authors Zeyu Ding, Katja Ickstadt, Nadja Klein, Alexander Munteanu, Simon Omlor

Machine Learning · Cross-listed from cs.CV

Growing Networks with Autonomous Pruning

This paper introduces Growing Networks with Autonomous Pruning (GNAP) for image classification. Unlike traditional convolutional neural networks, GNAP change their size, as well as the number of parameters they are using, during training, in order to best fit the data while trying to use as few parameters as possible. This is achieved through two complementary mechanisms: growth and pruning. GNAP start with few parameters, but their size is expanded periodically during training to add more expressive power each time the network has converged to a saturation point. Between these growing phases, model parameters are trained for classification and pruned simultaneously, with complete autonomy by gradient descent. Growing phases allow GNAP to improve their classification performance, while autonomous pruning allows them to keep as few parameters as possible. Experimental results on several image classification benchmarks show that our approach can train extremely sparse neural networks with high accuracy. For example, on MNIST, we achieved 99.44% accuracy with as few as 6.2k parameters, while on CIFAR10, we achieved 92.2\ accuracy with 157.8k parameters.
Read the full abstract

Machine Learning

Layer-wise Target Propagation: Efficient Component Attribution through Target Centric Propagation

Understanding the internal mechanisms of transformer-based large language models (LLMs) is crucial for their reliable deployment and effective operation. While recent efforts have yielded a plethora of attribution methods attempting to balance faithfulness and computational efficiency, dense component attribution remains prohibitively expensive. In this work, we introduce Layer-wise Target Propagation (LTP), a novel framework that faithfully traces information flow on the frozen transformer in one forward and one backward pass without requiring counterfactual examples. LTP analytically decomposes and linearizes the computational structure of the Transformers into distinct pathways along which it propagates a targeted unembedding vector to receive the effective representation at each residual position. This target-centric propagation achieves O(1) time complexity with respect to the number of model components, scaling to long input sequences and dense component attribution. Extensive experiments on standard interpretability benchmarks demonstrate that LTP achieves state-of-the-art faithfulness and unprecedented efficiency compared to existing baselines.
Read the full abstract

All authors Lasse Marten Jantsch, Dong-Jae Koh, Seonghyeon Lee, Young-Kyoon Suh

Machine Learning

FedPDPO: Federated Personalized Direct Preference Optimization for Large Language Model Alignment

Aligning large language models (LLMs) with human preferences in federated learning (FL) is challenging due to decentralized, privacy-sensitive, and highly non-IID preference data. Direct Preference Optimization (DPO) offers an efficient alternative to reinforcement learning with human feedback (RLHF), but its direct application in FL suffers from severe performance degradation under non-IID data and limited generalization of implicit rewards. To bridge this gap, we propose FedPDPO (Federated Personalized Direct Preference Optimization), a personalized federated framework for preference alignment of LLMs. It adopts a parameter-efficient fine-tuning architecture where each client maintains a frozen pretrained LLM backbone augmented with a Low-Rank Adaptation (LoRA) adapter, enabling communication-efficient aggregation. To address non-IID heterogeneity, we devise (1) the globally shared LoRA adapter with the personalized client-specific LLM head. Moreover, we introduce (2) a personalized DPO training strategy with a client-specific explicit reward head to complement implicit rewards and further alleviate non-IID heterogeneity, and (3) a bottleneck adapter to balance global and local features. We provide theoretical analysis establishing the probabilistic foundation and soundness. Extensive experiments on multiple preference datasets demonstrate state-of-the-art performance, achieving up to 4.80% average accuracy improvements in federated intra-domain and cross-domain settings.
Read the full abstract

All authors Kewen Zhu, Liping Yi, Zhiming Zhao, Zhuang Qi, Han Yu, Qinghua Hu

Machine Learning · Cross-listed from stat.ML

A two-step sequential approach for hyperparameter selection in finite context models

Finite-context models (FCMs) are widely used for compressing symbolic sequences such as DNA, where predictive performance depends critically on the context length k and smoothing parameter {\alpha}. In practice, these hyperparameters are typically selected through exhaustive search, which is computationally expensive and scales poorly with model complexity. This paper proposes a statistically grounded two-step sequential approach for efficient hyperparameter selection in FCMs. The key idea is to decompose the joint optimization problem into two independent stages. First, the context length k is estimated using categorical serial dependence measures, including Cramér's {\nu}, Cohen's ąppa} and partial mutual information (pami). Second, the smoothing parameter {\alpha} is estimated via maximum likelihood conditional on the selected context length k. Simulation experiments were conducted on synthetic symbolic sequences generated by FCMs across multiple (k, {\alpha}) configurations, considering a four-letter alphabet and different sample sizes. Results show that the dependence measures are substantially more sensitive to variations in k than in {\alpha}, supporting the sequential estimation strategy. As expected, the accuracy of the hyperparameter estimation improves with increasing sample size. Furthermore, the proposed method achieves compression performance comparable to exhaustive grid search in terms of average bitrate (bits per symbol), while substantially reducing computational cost. Overall, the results on simulated data show that the proposed sequential approach is a practical and computationally efficient alternative to exhaustive hyperparameter tuning in FCMs.
Read the full abstract

All authors José Contente, Ana Martins, Armando J. Pinho, Sónia Gouveia

Machine Learning

FedRG: Unleashing the Representation Geometry for Federated Learning with Noisy Clients

Federated learning (FL) suffers from performance degradation due to the inevitable presence of noisy annotations in distributed scenarios. Existing approaches have advanced in distinguishing noisy samples from the dataset for label correction by leveraging loss values. However, noisy samples recognition relying on scalar loss lacks reliability for FL under heterogeneous scenarios. In this paper, we rethink this paradigm from a representation perspective and propose \method~(Federated under Representation Gemometry), which follows the principle of “representation geometry priority” to recognize noisy labels. Firstly, \method~creates label-agnostic spherical representations by using self-supervision. It then iteratively fits a spherical von Mises-Fisher (vMF) mixture model to this geometry using previously identified clean samples to capture semantic clusters. This geometric evidence is integrated with a semantic-label soft mapping mechanism to derive a distribution divergence between the label-free and annotated label-conditioned feature space, which robustly identifies noisy samples and updates the vMF mixture model with the newly separated clean dataset. Lastly, we employ an additional personalized noise absorption matrix on noisy labels to achieve robust optimization. Extensive experimental results demonstrate that \method~significantly outperforms state-of-the-art methods for FL with data heterogeneity under diverse noisy clients scenarios.
Read the full abstract

All authors Tian Wen, Zhiqin Yang, Yonggang Zhang, Xuefeng Jiang, Hao Peng, Yuwei Wang, Bo Han

Machine Learning

Binary Classification from Coupled Pairwise Labels

Even when it is difficult to assign absolute class labels to individual instances, relational information may still be available, such as whether two instances belong to the same class or which instance is more likely to belong to the positive class. In this study, we refer to these two types of information as Similarity/Dissimilarity (SD) labels and Pairwise Comparison (Pcomp) labels, respectively, and consider binary classification that uses both types of relational information from the same instance pairs. SD learning uses the distinction between similar and dissimilar pairs but does not use the ordering within each pair, whereas Pcomp learning uses the ordering within each pair but does not distinguish between similar and dissimilar pairs. We therefore propose SD-Pcomp learning, whose objective function simultaneously preserves the structures of both SD learning and Pcomp learning. The proposed objective function admits two decompositions: one consists of an SD estimator plus a term that represents ordering information from Pcomp labels, and the other consists of a Pcomp estimator plus a term that represents pair-type information from SD labels. These decompositions clarify how the complementary information provided by SD and Pcomp labels is integrated into the proposed objective function. Experiments on eight datasets compare the proposed method with SD learning, Pcomp learning, and a method that takes a convex combination of their objective functions. We evaluate the effect of using both types of relational information on classification performance in terms of classification accuracy and AUC.
Read the full abstract

Machine Learning · Cross-listed from math.ST

Minimax and Adaptive Covariance Matrix Estimation under Differential Privacy

Estimating covariance matrices is fundamental to a wide range of statistical applications. This paper studies minimax and adaptive estimation of high-dimensional covariance matrices under $\rho$-zero-concentrated differential privacy ($\rho$-zCDP) over three nested classes: the pointwise-decay class $\mathcal{H}_\alpha$, the row-tail class $\mathcal{G}_\alpha$, and the separated-block class $\mathcal{F}_\alpha$. We consider both squared operator norm loss and normalized squared Frobenius norm loss. For $\mathcal{H}_\alpha$ and $\mathcal{G}_\alpha$, we develop center–outer dyadic estimators tailored to the refined geometry of the two classes, while for $\mathcal{F}_\alpha$, we develop a blockwise tridiagonal estimator. The resulting minimax-optimal rates reveal a nontrivial interplay among the smoothness $\alpha$, the loss, the geometry of the covariance class, and the privacy constraint. In contrast to the non-private setting, privacy distinguishes covariance classes that share the same leading non-private rate and induces a polynomial dependence on the ambient dimension. We further develop procedures that adapt to the unknown decay parameter over all three covariance classes under both losses, at the cost of at most polylogarithmic factors. To establish minimax lower bounds, we develop a novel differentially private van Trees inequality that connects Fisher information with the $\rho$-zCDP constraint and may be useful for other private estimation problems. We also construct carefully designed prior distributions to obtain matching minimax lower bounds.
Read the full abstract

Machine Learning

Regret Analysis of Sleeping Competing Bandits

The Competing Bandits framework is a recently emerging area that integrates multi-armed bandits in online learning with stable matching in game theory. While conventional models assume that all players and arms are constantly available, in real-world problems, their availability can vary arbitrarily over time. In this paper, we formulate this setting as Sleeping Competing Bandits. To analyze this problem, we naturally extend the regret definition used in existing competing bandits and derive regret bounds for the proposed model. We propose an algorithm that simultaneously achieves an asymptotic regret bound of $\mathrm{O}\left(NK\log T_{i}/\Delta^2\right)$ under reasonable assumptions, where $N$ is the number of players, $K$ is the number of arms, $T_{i}$ is the number of rounds of each player $p_i$, and $\Delta$ is the minimum reward gap. We also provide a regret lower bound of $\mathrm{\Omega}\left( N(K-N+1)\log T_{i}/\Delta^2 \right)$ under the same assumptions. This implies that our algorithm is asymptotically optimal in the regime where the number of arms $K$ is relatively larger than the number of players $N$.
Read the full abstract

Machine Learning · Cross-listed from cs.LO

Diminishing Returns in Expanding Generative Models and Godel-Tarski-Lob Limits

Modern generative modelling systems are increasingly improved by expanding model capacity, training data, and computational resources. While empirical studies have documented such scaling behaviour across architectures including generative adversarial networks, variational autoencoders, transformer-based models, and diffusion models, the theoretical limits of capability growth in expanding generative systems remain poorly understood. In this paper we develop a general task-space framework for analysing expanding generative reasoning systems. Each system induces a subset of a global task space representing the tasks it can successfully solve, and system capability is measured by the probability mass of this solved-task set under a fixed task distribution. Within this framework we prove a structural result showing that, under mild assumptions, the marginal improvement in solved tasks must converge to zero as system capacity increases. Thus expanding generative systems may continue to gain capability, but the probability mass of newly solvable tasks necessarily diminishes asymptotically. We further provide a prediction-theoretic refinement based on complexity-weighted hypothesis classes inspired by algorithmic probability, yielding quantitative bounds on marginal improvement in prediction settings. Finally, we examine logical reasoning tasks and show that classical results from mathematical logic – including Rosser incompleteness, Tarski's undefinability theorem, and Löb's theorem – imply the persistence of unresolved logical tasks within sufficiently expressive reasoning systems. Together these results provide a mathematical perspective on the asymptotic behaviour of expanding generative systems, showing that long-run capability growth is constrained both by diminishing marginal improvements in task coverage and by fundamental logical limitations on internal reasoning.
Read the full abstract

Machine Learning · Cross-listed from cs.AI

A Subgoal-driven Framework for Improving Long-Horizon LLM Agents

Large language model (LLM)-based agents have emerged as powerful autonomous controllers for digital environments, including mobile interfaces, operating systems, and web browsers. Web navigation, for example, requires handling dynamic content and long sequences of actions, making it particularly challenging. Existing LLM-based agents struggle with long-horizon planning in two main ways. During online execution, they often lose track as new information arrives, lacking a clear and adaptive path toward the final goal. This issue is further exacerbated during reinforcement learning (RL) fine-tuning, where sparse and delayed rewards make it difficult for agents to identify which actions lead to success, preventing them from maintaining coherent reasoning over extended tasks. To address these challenges, we propose two contributions. First, we introduce an agent framework that leverages proprietary models for online planning through subgoal decomposition. Second, we present MiRA (Milestoning your Reinforcement Learning Enhanced Agent), an RL training framework that uses dense, milestone-based reward signals. The real-time planning mechanism improves proprietary models such as Gemini by approximately a 10% absolute increase in success rate (SR) on the WebArena-Lite benchmark. Meanwhile, applying MiRA to the open Gemma3-12B model increases its success rate from 6.4% to 43.0%. This performance surpasses proprietary systems such as GPT-4-Turbo (17.6%) and GPT-4o (13.9%), as well as the previous open-model state of the art, WebRL (38.4%). Overall, our findings demonstrate that combining explicit inference-time planning with milestone-based rewards significantly improves an agent's long-horizon capabilities, paving the way for more robust and general-purpose autonomous systems.
Read the full abstract

All authors Taiyi Wang, Sian Gooding, Florian Hartmann, Oriana Riva, Edward Grefenstette

Machine Learning

Ontology-Based Knowledge Modeling and Uncertainty-Aware Outdoor Air Quality Assessment Using Weighted Interval Type-2 Fuzzy Logic

Outdoor air pollution is a major concern for the environment and public health, especially in areas where urbanization is taking place rapidly. The Indian Air Quality Index (IND-AQI), developed by the Central Pollution Control Board (CPCB), is a standardized reporting system for air quality based on pollutants such as PM2.5, PM10), nitrogen dioxide (NO2), sulfur dioxide (SO2), ozone (O3), carbon monoxide (CO), and ammonia (NH3). However, the traditional calculation of the AQI uses crisp thresholds and deterministic aggregation rules, which are not suitable for handling uncertainty and transitions between classes. To address these limitations, this study proposes a hybrid ontology-based uncertainty-aware framework integrating Weighted Interval Type-2 Fuzzy Logic with semantic knowledge modeling. Interval Type-2 fuzzy sets are used to model uncertainty near AQI class boundaries, while pollutant importance weights are determined using Interval Type-2 Fuzzy Analytic Hierarchy Process (IT2-FAHP) to reflect their relative health impacts. In addition, an OWL-based air quality ontology extending the Semantic Sensor Network (SSN) ontology is developed to represent pollutants, monitoring stations, AQI categories, regulatory standards, and environmental governance actions. Semantic reasoning is implemented using SWRL rules and validated through SPARQL queries to infer AQI categories, health risks, and recommended mitigation actions. Experimental evaluation using CPCB air quality datasets demonstrates that the proposed framework improves AQI classification reliability and uncertainty handling compared with traditional crisp and Type-1 fuzzy approaches, while enabling explainable semantic reasoning and intelligent decision support for air quality monitoring systems
Read the full abstract

All authors Md Inzmam, Ritesh Chandra, Sadhana Tiwari, Sonali Agarwal, Triloki Pant

Machine Learning

GoAgent: Group-of-Agents Communication Topology Generation for LLM-based Multi-Agent Systems

Large language model (LLM)-based multi-agent systems (MAS) have demonstrated exceptional capabilities in solving complex tasks, yet their effectiveness depends heavily on the underlying communication topology that coordinates agent interactions. Within these systems, successful problem-solving often necessitates task-specific group structures to divide and conquer subtasks. However, most existing approaches generate communication topologies in a node-centric manner, leaving group structures to emerge implicitly from local connectivity decisions rather than modeling them explicitly, often leading to suboptimal coordination and unnecessary communication overhead. To address this limitation, we propose GoAgent (Group-of-Agents), a communication topology generation method that explicitly treats collaborative groups as the atomic units of MAS construction. Specifically, GoAgent first enumerates task-relevant candidate groups through an LLM and then autoregressively selects and connects these groups as atomic units to construct the final communication graph, jointly capturing intra-group cohesion and inter-group coordination. To mitigate communication redundancy and noise propagation inherent in expanding topologies, we further introduce a conditional information bottleneck (CIB) objective that compresses inter-group communication, preserving task-relevant signals while filtering out redundant historical noise. Extensive experiments on six benchmarks demonstrate the state-of-the-art performance of GoAgent with 93.84% average accuracy while reducing token consumption by about 17%.
Read the full abstract

All authors Hongjiang Chen, Xin Zheng, Yixin Liu, Pengfei Jiao, Shiyuan Li, Huan Liu, Zhidong Zhao, Ziqi Xu, Ibrahim Khalil, Shirui Pan

Machine Learning · Cross-listed from cs.CV

ATHENA: Adaptive Test-Time Steering for Improving Count Fidelity in Diffusion Models

Text-to-image diffusion models achieve high visual fidelity but surprisingly exhibit systematic failures in numerical control when prompts specify explicit object counts. To address this limitation, we introduce ATHENA, a model-agnostic, test-time adaptive steering framework that improves object count fidelity without modifying model architectures or requiring retraining. ATHENA leverages intermediate representations during sampling to estimate object counts and applies count-aware noise corrections early in the denoising process, steering the generation trajectory before structural errors become difficult to revise. We present three progressively more advanced variants of ATHENA that trade additional computation for improved numerical accuracy, ranging from static prompt-based steering to dynamically adjusted count-aware control. Experiments on established benchmarks and a new visually and semantically complex dataset show that ATHENA consistently improves count fidelity, particularly at higher target counts, while maintaining favorable accuracy-runtime trade-offs across multiple diffusion backbones.
Read the full abstract

All authors Mohammad Shahab Sepehri, Asal Mehradfar, Berk Tinaz, Salman Avestimehr, Mahdi Soltanolkotabi

Machine Learning

Load–Reserve Wasserstein Propagation for Isotropic Diffusion Samplers

Many Wasserstein analyses of diffusion samplers control reverse-time propagation by global stability summaries of the learned drift. These summaries can hide radial geometry: equal-height expansive regions of different width can yield different propagation costs. We give a profile-adapted propagation interface for scalar-isotropic reverse-SDE windows with certified learned-drift profiles. A certified lower radial profile is compiled into an affine-tail transportation cost: reflection coupling reduces stability to a one-dimensional slope budget, and Hardy capacity quantifies the load paid before a contractive tail reserve. The compiler yields an adapted cost, contraction rate, and retained tail slope. Score-modeling and solver residuals are treated as forcing inputs and propagate additively in the adapted Wasserstein distance. Quadratic Wasserstein error is reported only at terminal time, using the retained tail slope with tail, moment, or support information. Gaussian-smoothed denoising geometry supplies inverse-radius profiles for uniformly dissipative, bounded-amplitude, and common-covariance mixture windows. Fixed-height examples show that adverse height, even with eventual reserve, does not determine the certificate; barrier examples show that the load dependence is structural.
Read the full abstract

Machine Learning

The Residual Stream Is All You Need: On the Redundancy of the KV Cache in Transformer Inference

The key-value (KV) cache is widely treated as essential state in transformer inference, and a large body of work engineers policies to compress, evict, or approximate its entries. We prove that this state is entirely redundant: keys and values at every layer are deterministic projections of the residual stream, and recomputing them from a single residual vector per token incurs exactly zero reconstruction error, not approximately, but bit-identically. We verify this across six models from four architecture families (135M to 4B parameters). Cross-task residual patching at every layer produces D_KL = 0 between patched and original output distributions, confirming that the residual stream satisfies a Markov property and is the sole information-carrying state. Removing the cache entirely and recomputing from scratch yields token-identical output under greedy decoding on all models tested. We build on this result with KV-Direct, a bounded-memory inference scheme that checkpoints residual vectors (5 KB per token on Gemma 3-4B) instead of full KV pairs (136 KB), recomputing keys and values on demand. Over 20 conversation turns, KV-Direct holds peak memory at 42 MB while the standard cache grows past 103 MB. Against five eviction baselines (H2O, StreamingLLM, SnapKV, TOVA, window-only), KV-Direct maintains 100% token match at every cache budget; all baselines degrade to 5-28%. A per-operation latency analysis shows recomputation runs up to 5x faster than reading cached tensors at moderate batch sizes. Code is available at https://github.com/Kaleemullahqasim/KV-Direct.
Read the full abstract

All authors Kaleem Ullah Qasim, Jiashu Zhang, Muhammad Kafeel Shaheen, Razan Alharith, Heying Zhang

Machine Learning · Cross-listed from stat.ML

Model Selection and Parameter Estimation for Multidimensional Gaussian Mixture Models with a Common Covariance Matrix

We study model-order selection and component-mean estimation for multidimensional Gaussian mixture models with a known common covariance matrix. Using empirical characteristic-function measurements, we construct Fourier covariance matrices whose population counterparts have rank equal to the number of mixture components. We establish a minimax lower bound showing that distinguishing a separated $k$-component mixture from the class of $(k-1)$-component mixtures requires $\Omega(\Delta^{-(4k-4)})$ samples. We then develop an oracle spectral-thresholding estimator with a sufficient sample size of order $\Delta^{-(8k-8)}$ for fixed $k$, together with a practical singular-value-ratio estimator. Given the model order, we estimate the component means by score-initialized gradient descent on a MUSIC-type projection objective. Under an explicit sample-size condition, a qualifying sample initialization lies in a certified attraction region with high probability, after which the iterates converge linearly. For fixed positive component separation, the resulting mean estimates achieve the parametric rate $\mathcal{O}_p(n^{-1/2})$. Numerical experiments demonstrate competitive accuracy and lower computational cost than expectation-maximization across a range of multidimensional settings.
Read the full abstract

Machine Learning

Ensembles-based Feature Guided Analysis

Recent Deep Neural Networks (DNN) applications ask for techniques that can explain their behavior. Existing solutions, such as Feature Guided Analysis (FGA), extract rules on their internal behaviors, e.g., by providing explanations related to neurons activation. Results from the literature show that these rules have considerable precision (i.e., they correctly predict certain classes of features), but the recall (i.e., the number of situations these rule apply) is more limited. To mitigate this problem, this paper presents Ensembles-based Feature Guided Analysis (EFGA). EFGA combines rules extracted by FGA into ensembles. Ensembles aggregate different rules to increase their applicability depending on an aggregation criterion, a policy that dictates how to combine rules into ensembles. Although our solution is extensible, and different aggregation criteria can be developed by users, in this work, we considered three different aggregation criteria. We evaluated how the choice of the criterion influences the effectiveness of EFGA on two benchmarks (i.e., the MNIST and LSC datasets), and found that different aggregation criteria offer alternative trade-offs between precision and recall. We then compare EFGA with FGA. For this experiment, we selected an aggregation criterion that provides a reasonable trade-off between precision and recall. Our results show that EFGA has higher train recall (+28.51% on MNIST, +33.15% on LSC), and test recall (+25.76% on MNIST, +30.81% on LSC) than FGA, with a negligible reduction on the test precision (-0.89% on MNIST, -0.69% on LSC).
Read the full abstract

All authors Federico Formica, Stefano Gregis, Andrea Rota, Aurora Francesca Zanenga, Mark Lawford, Claudio Menghi

Machine Learning

Heavy-Tailed and Long-Range Dependent Noise in Stochastic Approximation: A Finite-Time Analysis

Stochastic approximation (SA) is a fundamental iterative framework with broad applications in reinforcement learning and optimization. Classical analyses typically rely on martingale difference or Markov noise with bounded second moments, but many practical settings, including finance and communications, frequently encounter heavy-tailed and long-range dependent (LRD) noise. In this work, we study SA for finding the root of a strongly monotone operator under these non-classical noise models. We establish the first finite-time moment bounds in both settings, providing explicit convergence rates that quantify the impact of heavy tails and temporal dependence. Our analysis employs a noise-averaging argument that regularizes the impact of noise without modifying the iteration. Finally, we apply our general framework to stochastic gradient descent (SGD) and gradient play, and corroborate our finite-time analysis through numerical experiments.
Read the full abstract

All authors Siddharth Chandak, Anuj Yadav, Ayfer Ozgur, Nicholas Bambos

Machine Learning

RiboSphere: Learning Unified and Efficient Representations of RNA Structures

Accurate RNA structure modeling remains difficult because RNA backbones are highly flexible, non-canonical interactions are prevalent, and experimentally determined 3D structures are comparatively scarce. We introduce RiboSphere, a framework that learns discrete geometric representations of RNA by combining vector quantization with flow matching. Our design is motivated by the modular organization of RNA architecture: complex folds are composed from recurring structural motifs. RiboSphere uses a geometric transformer encoder trained using mean-centered coordinates and random rotation augmentation to produce geometry-aware features, which are discretized with finite scalar quantization (FSQ) into a finite vocabulary of latent codes. Conditioned on these discrete codes, a flow-matching decoder reconstructs atomic coordinates, enabling high-fidelity structure generation. We find that the learned code indices are enriched for specific RNA motifs, suggesting that the model captures motif-level compositional structure rather than acting as a purely compressive bottleneck. Across benchmarks, RiboSphere achieves strong performance in structure reconstruction (RMSD 1.25,Å, TM-score 0.84), and its pretrained discrete representations transfer effectively to inverse folding and RNA–ligand binding prediction, with robust generalization in data-scarce regimes. Code is available at https://github.com/Zhangz312/RiboSphere.
Read the full abstract

All authors Zhou Zhang, Hanqun Cao, Cheng Tan, Fang Wu, Pheng Ann Heng, Tianfan Fu

Machine Learning

Alternating Diffusion for Proximal Sampling with Zeroth Order Queries

This work introduces a new approximate proximal sampler that operates solely with zeroth-order information of the potential function. Prior theoretical analyses have revealed that proximal sampling corresponds to alternating forward and backward iterations of the heat flow. The backward step was originally implemented by rejection sampling, whereas we directly simulate the dynamics. Unlike diffusion-based sampling methods that estimate scores via learned models or by invoking auxiliary samplers, our method treats the intermediate particle distribution as a Gaussian mixture, thereby yielding a Monte Carlo score estimator from directly samplable distributions. Theoretically, when the score estimation error is sufficiently controlled, our method inherits the exponential convergence of proximal sampling under isoperimetric conditions on the target distribution. In practice, the algorithm avoids rejection sampling, permits flexible step sizes, and runs with a deterministic runtime budget. Numerical experiments demonstrate that our approach converges rapidly to the target distribution, driven by interactions among multiple particles and by exploiting parallel computation.
Read the full abstract

Machine Learning · Cross-listed from stat.ML

On the role of memorization in learned priors for geophysical inverse problems

Learned priors based on deep generative models offer data-driven regularization for seismic inversion, but training them requires a dataset of representative subsurface models – a resource that is inherently scarce in geoscience applications. Since the training objective of most generative models can be cast as maximum likelihood on a finite dataset, any such model risks converging to the empirical distribution – effectively memorizing the training examples rather than learning the underlying geological distribution. We show that the posterior under such a memorized prior reduces to a reweighted empirical distribution – i.e., a likelihood-weighted lookup among the stored training examples. For diffusion models specifically, memorization yields a Gaussian mixture prior in closed form, and linearizing the forward operator around each training example gives a Gaussian mixture posterior whose components have widths and shifts governed by the local Jacobian. We validate these predictions on a stylized inverse problem and demonstrate the consequences of memorization through diffusion posterior sampling for full waveform inversion.
Read the full abstract

Machine Learning

Continual Learning for Food Category Classification Dataset: Enhancing Model Adaptability and Performance

Conventional machine learning pipelines often struggle to recognize categories absent from the original trainingset. This gap typically reduces accuracy, as fixed datasets rarely capture the full diversity of a domain. To address this, we propose a continual learning framework for text-guided food classification. Unlike approaches that require retraining from scratch, our method enables incremental updates, allowing new categories to be integrated without degrading prior knowledge. For example, a model trained on Western cuisines could later learn to classify dishes such as dosa or kimchi. Although further refinements are needed, this design shows promise for adaptive food recognition, with applications in dietary monitoring and personalized nutrition planning.
Read the full abstract

All authors Piyush Kaushik Bhattacharyya, Devansh Tomar, Shubham Mishra, Divyanshu Rai, Yug Pratap Singh, Harsh Yadav, Krutika Verma, Vishal Meena, N Sangita Achary

Machine Learning

DeepStock: Reinforcement Learning with Policy Regularizations for Inventory Management

Deep Reinforcement Learning (DRL) provides a general-purpose methodology for training inventory policies that can leverage big data and compute. However, off-the-shelf implementations of DRL have seen mixed success, often plagued by high sensitivity to the hyperparameters used during training. In this paper, we show that by imposing policy regularizations, grounded in classical inventory concepts such as "Base Stock", we can significantly accelerate hyperparameter tuning and improve the final performance of several DRL methods. We report details from a 100% deployment of DRL with policy regularizations on Alibaba's e-commerce platform, Tmall. We also include extensive synthetic experiments, which show that policy regularizations reshape the narrative on what is the best DRL method for inventory management.
Read the full abstract

All authors Yaqi Xie, Xinru Hao, Jiaxi Liu, Will Ma, Linwei Xin, Lei Cao, Yidong Zhang

Machine Learning

A Locally Penalized Cross-Estimate Federated Method with Guarantees for Constrained Personalized Learning

Federated learning (FL) is a communication-efficient framework for solving distributed optimization problems. Standard FL methods aggregate local updates into a single global model that is shared by all agents, effectively imposing consensus. Under heterogeneous local constraints, however, a common feasible model may be overly restrictive or may not exist. Existing constrained FL methods largely retain this shared-model structure and therefore do not directly address personalization under heterogeneous agent-specific feasible sets. We study a constrained personalized FL problem that assigns a distinct feasible model to each agent while coupling the models through a collaborative objective. We propose Locally Penalized Cross-Estimate Federated Averaging (PCE-FedAvg), where each agent maintains a multi-block vector containing its own model and estimates of the other agents' models while applying the feasibility penalty only locally to its own block. The server aggregates corresponding blocks separately, thereby preserving personalization without requiring agents to explicitly disclose their local constraint sets. For any $\epsilon>0$, we establish finite-time upper and lower suboptimality bounds and agent-wise squared infeasibility bounds, yielding communication complexities of $\mathcal{O}(\epsilon^{-2})$ and $\mathcal{O}(\epsilon^{-1})$, respectively. Experiments on MNIST and CIFAR-10 support our theoretical results.
Read the full abstract

Machine Learning

Demonstrations, CoT, and Prompting: A Theoretical Analysis of ICL

In-Context Learning (ICL) enables pretrained LLMs to adapt to downstream tasks by conditioning on a small set of input-output demonstrations, without any parameter updates. Although there have been many theoretical efforts to explain how ICL works, most either rely on strong architectural or data assumptions, or fail to capture the impact of key practical factors such as demonstration selection, Chain-of-Thought (CoT) prompting, the number of demonstrations, and prompt templates. We address this gap by establishing a theoretical analysis of ICL under mild assumptions that links these design choices to generalization behavior. We derive an upper bound on the ICL test loss, showing that performance is governed by (i) the quality of selected demonstrations, quantified by Lipschitz constants of the ICL loss along paths connecting test prompts to pretraining samples, (ii) an intrinsic ICL capability of the pretrained model, and (iii) the degree of distribution shift. Within the same framework, we analyze CoT prompting as inducing a task decomposition and show that it is beneficial when demonstrations are well chosen at each substep and the resulting subtasks are easier to learn. Finally, we characterize how ICL performance sensitivity to prompt templates varies with the number of demonstrations. Together, our study shows that pretraining equips the model with the ability to generalize beyond observed tasks, while CoT enables the model to compose simpler subtasks into more complex ones, and demonstrations and instructions enable it to retrieve similar or complex tasks, including those that can be composed into more complex ones, jointly supporting generalization to unseen tasks. All theoretical insights are corroborated by experiments.
Read the full abstract

Machine Learning · Cross-listed from cs.CV

K-GMRF: Kinetic Gauss-Markov Random Field for First-Principles Covariance Tracking on Lie Groups

Tracking non-stationary covariance matrices is fundamental to vision yet hindered by existing estimators that either neglect manifold constraints or rely on first-order updates, incurring inevitable phase lag during rapid evolution. We propose K-GMRF, an online, training-free framework for covariance tracking that reformulates the problem as forced rigid-body motion on Lie groups. Derived from the Euler-Poincaré equations, our method interprets observations as torques driving a latent angular velocity, propagated via a structure-preserving symplectic integrator. We theoretically prove that this second-order dynamics achieves zero steady-state error under constant rotation, strictly superior to the proportional lag of first-order baselines. Validation across three domains demonstrates robust tracking fidelity: (i) on synthetic ellipses, K-GMRF reduces angular error by 30x compared to Riemannian EMA while maintaining stability at high speeds; (ii) on SO(3) stabilization with 20% dropout, it decreases geodesic error from 29.4{\deg} to 9.9{\deg}; and (iii) on OTB motion-blur sequences, it improves loU from 0.55 to 0.74 on BlurCar2 with a 96% success rate. As a fully differentiable symplectic module, K-GMRF provides a plug-and-play geometric prior for data-constrained scenarios and an interpretable layer within modern deep architectures.
Read the full abstract

Machine Learning

ARMOR: Adaptive Resilience Against Model Poisoning Attacks in Continual Federated Learning for Mobile Indoor Localization

Indoor localization has become increasingly essential for applications ranging from asset tracking to delivering personalized services. Federated learning (FL) offers a privacy-preserving approach by training a centralized global model (GM) using distributed data from mobile devices without sharing raw data. However, real-world deployments require a continual federated learning (CFL) setting, where the GM receives continual updates under device heterogeneity and evolving indoor environments. In such dynamic conditions, erroneous or biased updates can cause the GM to deviate from its expected learning trajectory, gradually degrading internal GM representations and GM localization performance. This vulnerability is further exacerbated by adversarial model poisoning attacks. To address this challenge, we propose ARMOR, a novel CFL-based framework that monitors and safeguards the GM during continual updates. ARMOR introduces a novel state-space model (SSM) that learns the historical evolution of GM weight tensors and predicts the expected next state of weight tensors of the GM. By comparing incoming local updates with this SSM projection, ARMOR detects deviations and selectively mitigates corrupted updates before local updates are aggregated with the GM. This mechanism enables robust adaptation to temporal environmental dynamics and mitigate the effects of model poisoning attacks while preventing GM corruption. Experimental evaluations in real-world conditions indicate that ARMOR achieves notable improvements, with up to 8.0x reduction in mean error and 4.97x reduction in worst-case error compared to state-of-the-art indoor localization frameworks, demonstrating strong resilience against model corruption tested using real-world data and mobile devices.
Read the full abstract