Papers

Mathematical search problems present a unique challenge for Reinforcement Learning (RL) due to vast search spaces and sparse rewards. In previous works, the Andrews-Curtis (AC) conjecture was established as an illustrative example of such problems. In this work, we identify a critical structural barrier in the AC landscape: a "Two-Hump" distribution, where problem instances are either trivially solvable or effectively impossible, with a scarcity of intermediate "hard-but-solvable" instances required for effective learning. We tackle this challenge through two primary avenues: novel data generation techniques to populate the difficulty gap, and significant algorithmic enhancements including the introduction of supermoves and Transformer-based architectures. We demonstrate substantial performance improvements over previous baselines, and release new comprehensive benchmark datasets including AC-19 (125,192 AC-trivial presentations of varying difficulty with length at most 19) and AC-1M (1,136,154 hard AC-trivial presentations of length at most 30), the first large-scale, publicly available datasets of this kind.

Neural combinatorial optimization (NCO) has shown that policies trained by reinforcement can construct strong solutions to NP-hard problems directly from raw instances. What such a policy actually learns, as opposed to what its decoder expresses, remains much less clear. We study this distinction on the vertex-guard Art Gallery Problem, the NP-hard task of choosing polygon vertices from which to observe an entire region. A pointer-network policy is trained from a coverage-aware reward over its own rollouts under the constraint we call geo-free inference: at test time it sees only vertex coordinates, with no visibility computation and no geometric oracle. The policy places guards economically but leaves a tail of under-covered polygons that widens far beyond the training range. To locate the cause, we freeze the trained encoder and read its embeddings with a small single-shot classifier, still geo-free at inference. The classifier closes most of the feasibility gap, in and out of distribution and at up to roughly five times the training range, cutting under-covered polygons by about an order of magnitude at an explicitly reported cost in guard count. We read this as evidence that the reinforcement-trained representation already encodes the geometry required for feasibility, and that residual failures reflect decoder calibration rather than missing knowledge. Probing a frozen encoder thus offers a practical way to ask what a neural combinatorial solver has internalized.

Deep neural networks transform input data into latent representations that support a wide range of downstream tasks. These representations can be characterized along information-theoretic and geometric dimensions, but their relationship remains poorly understood. A central open question is whether low mutual information (MI) between inputs and representations necessarily implies geometrically compressed latent spaces and vice versa. We investigate this question using class-wise clustering as a measure of geometric compression and theoretically sound MI estimation in conditional entropy bottleneck (CEB) networks and continuous dropout networks. We evaluate the interplay between MI, geometric compression, and generalization on classification tasks under controlled noise injection schemes. Our findings show that low MI does not reliably correspond to geometric compression, and that the connection between the two is more nuanced than often assumed. Indeed, our experiments reveal a negative and nonlinear relationship that can reverse when varying training setup. Our results put forward a hypothesis that generalization acts as a potential confounder in this connection rather than being their direct consequence.

Self-supervised learning (SSL) typically relies on a backbone encoder followed by a small multilayer perceptron (MLP) projection head, which is conventionally discarded after training, while backbone quality is assessed via costly linear probing on labeled data. We argue that this approach including discarding the projector is rather computationally wasteful. Instead, we propose replacing the MLP head with a radial basis function network (RBFN), whose interpretable center and shape parameters can be exploited to judge representation quality without labels or a separate classifier. To this end, we introduce Scale-Normalized Separation (SNS), a novel label-free quality metric derived solely from the kernel centers and shapes learned during training. Across five canonical SSL architectures (MoCo, SimCLR, BYOL, SwAV and SimSiam) and four image classification datasets, we show that RBFN projection heads are competitive drop-in replacements for standard MLP projectors. We recommend constructing them with three RBF layers activated by the Gaussian radial basis function. Moreover, SNS exhibits strong to very strong positive correlation with established logistic regression metrics, demonstrating that a trained RBFN projector can act as a reliable proxy for backbone representation quality. We additionally publish a novel PyTorch compatible image classification dataset based on Google's Open Images V7 to facilitate reproducible research into representation learning.

Deep reinforcement learning is pivotal for closed-loop autonomous driving yet remains constrained by severe bottlenecks in sampling efficiency. Standard parallel sampling mitigates this but suffers from the straggler effect, where the premature termination of a single environment necessitates a synchronized batch re-initialization, leading to suboptimal sample utilization and prohibitive re-initialization latency. To address this, we propose FAST, a synchronous parallel framework tailored for closed-loop simulation. Specifically, FAST employs Dynamic Parallel Sampling Alignment (DPSA) to maintain vectorization synchronization by extending terminated episodes via virtual continuation, thereby decoupling the sampling loop from individual terminations. By dynamically triggering global truncation based on the termination rate of parallel clips, FAST effectively eliminates the bottleneck of premature resets without sacrificing data diversity. Furthermore, to strictly preserve theoretical consistency, we incorporate a Scaled Mask-Padding Optimization (SMPO) that leverages validity masking and adaptive loss normalization to nullify the bias from auxiliary padding data. Empirical evaluations demonstrate that FAST achieves at least a 1.78 times wall-clock speedup over the single-clip baseline while preserving statistical unbiasedness.

A finite agent, a machine's digital twin or any bounded reasoner, infers a fixed and noisy world through finite sensors, so its coherent output is a belief: a probability density over states (the Bayes posterior). Such an agent stops short of certainty, and revising a belief carries a cost. We propose a framework for belief costs based on optimal transport, motivated by these facts. We pose two postulates. P0 (the arena): a revision cost is a scalar price on optimal transport, so beliefs live in Wasserstein space. P1 (uniform pricing): one nat of knowledge costs the same metric length everywhere, the eikonal condition. Among conceivable pricing rules we study this one. Under P0 and P1 the cost metric is optimal transport conformally reweighted by Fisher information, $\tilde g_{e,U}=2(e+U)\,g_{W_2}$, and the Fisher family is a characterization: among continuous reliefs, uniform pricing is equivalent to $U=cJ$. Two consequences follow on the conformal class. Certainty sits at infinite cost-distance once the relief dominates the Fisher information, so a well-posed inference has a cost floor diverging at certainty (necessity conjectural beyond power laws). On location-scale leaves the geometry is hyperbolic, and the Stam bound places the Gaussian as the most curved one (at $e=0$). The results are geometric, in nats, and hold up to units: a change of cost unit rescales all distances and preserves every conclusion (boundary, eikonal family, hyperbolicity, Gaussian extremum), a gauge theorem; a global change of state units at $e=0$ is an isometry; the content lies in signs, rankings and ratios. Via Landauer (one nat worth $k_BT$) the cost floor becomes an energy floor: revising toward certainty would demand unbounded energy. Physics anchors the unit and enters no theorem. Removing either postulate leaves the selection open.

Transformers achieve strong performance, but their internal computations remain opaque. We view each Transformer layer as a dynamic graph whose nodes are token representations and per-head attention outputs, with Multi-Head Attention (ATT) and MLP as module boundaries. On this graph we use LIG (Layer-wise Integrated Gradients), which applies set-to-set Integrated Gradients (IG) at nonlinear module boundaries. Set-to-set IG applies IG to a map from a set of input token representations to a set of output representations, evaluating token-to-token contributions, which is not standard in prior IG applications. This extends IG from the usual scalar-objective setting to set-to-set maps via an L2 scalarization, and composes within-layer contributions in the spirit of Layer-wise Relevance Propagation (LRP), with IG completeness playing the role of LRP-style conservation at each boundary. We use LIG to analyze (i) the agreement between module-wise composition and layer-whole attribution under an L2 criterion, and (ii) within-layer information flow by tracing separated ATT and MLP contributions. On BERT-base and PTB, configurations that best preserved within-layer consistency used the target token's embedding as the ATT baseline and either the ATT output at a=0 or Zero as the MLP baseline. We therefore present LIG as a diagnostic XAI tool at module-boundary granularity, without model-specific retraining or per-operation interpreter design. Code is available at https://github.com/eightsuzuki/layer-wise-integrated-gradients.

Transformers are AI's workhorse but their computational cost becomes prohibitive when processing long sequences. We target long-horizon streaming vision and robotics applications, where it is particularly impractical to store and maintain a history of observations. Recurrent Transformers address this limitation by maintaining fixed-size memory but their performance lags behind that of transformers operating over the full observation history. We argue that this gap does not stem from architectural limitations, but from differences in how these models learn to compress past information. Without access to an observation history, recurrent models must explicitly decide what to retain in memory at each step, a significantly harder learning problem. In this work, we propose a distillation approach that transfers the compression strategy of a classical full-history transformer to a recurrent variant. We enable this by designing a teacher model that explicitly compresses its observation history into a fixed-size bottleneck representation and directly supervise the student's memory with this bottleneck representation, effectively aligning the two compression mechanisms. We show that this approach allows to train a recurrent latent robotic memory with linear-time complexity on the Mem-RPE task while substantially narrowing the performance gap to full-history transformers. We additionally validate the same principle on streaming visual question answering (VQA) and observe improved recurrent predictions thanks to memory distillation

Graph convolutional networks (GCNs) have demonstrated significant success in capturing complex user-item relationships for collaborative filtering (CF). However, due to their reliance on extensive model training, training-free graph filtering (GF)-based CF methods have emerged as a promising alternative, offering computational efficiency by smoothing graph signals via matrix operations. In particular, polynomial GF-based approaches demonstrate improved accuracy through their ability to design more expressive and flexible filtering functions. Despite these advantages, existing GF methods suffer from a critical memory bottleneck: they necessitate storing the full item similarity graph, incurring prohibitive memory costs for large-scale datasets, which limits their practical applicability. To tackle this challenge, we propose Mem-GF (Memory-efficient GF), a new GF-based CF method that departs from conventional designs by principally leveraging the structure of Krylov subspaces as a core mechanism for approximating polynomial graph filters without explicitly storing the item similarity graph. We theoretically analyze the minimum Krylov subspace size that guarantees lossless approximation. Through extensive experiments, we demonstrate that Mem-GF achieves up to 5.74$\times$ lower memory usage and 4.38$\times$ speedup in runtime, while consistently exceeding the recommendation accuracy of state-of-the-art GF and GCN-based methods. Mem-GF robustly scales to datasets with tens of millions of interactions, establishing itself as a practically viable and theoretically grounded solution for efficient CF.

We study first-order methods for solving monotone variational inequalities arising in min-max optimization. Classical approaches such as the extragradient method rely on two gradient queries per iteration, which limits their analysis and applicability in the online and stochastic settings. We propose a family of Generalized Optimistic Methods with Anchoring (GOMA), which combine two-time-scale optimistic updates with an anchoring term inspired by Halpern iteration. In the deterministic setting, GOMA achieves the optimal accelerated last-iterate rate $O(1/k^2)$ on the squared gradient norm for monotone Lipschitz operators. In the stochastic setting with unbounded variance, a simplified single-call variant of GOMA achieves a last-iterate convergence rate of $O(1/\sqrt{k})$ on the squared gradient norm. To the best of our knowledge, this is the first such guarantee for stochastic monotone Lipschitz variational inequalities in the unconstrained setting without variance reduction or growing batches.

Model-free reinforcement learning algorithms such as Proximal Policy Optimization (PPO) treat the environment as a black box, estimating policy gradients from sampled rewards; this process demands millions of interactions and relies on high-variance advantage estimates. When environment dynamics are differentiable, the return is an end-to-end differentiable function of the policy parameters, enabling exact gradient computation via backpropagation through simulation. We term this approach Analytic Policy Gradients (APG) and evaluate it against PPO on four continuous control tasks of increasing dynamical complexity: a one-dimensional point-mass target-reaching task, a 2D point-mass navigation task with obstacle avoidance, a 2D rigid-body T-block pushing task, and a 7-DOF Franka FR3 end-effector reaching task. Both algorithms share identical model architectures, observation normalization, and optimizer settings. To decouple sample efficiency from compute efficiency, we design a multi-axis evaluation protocol that records performance against environment steps and gradient steps. We report a segmented backpropagation scheme with MC and critic-based bootstrap modes that mitigates gradient degradation on long-horizon tasks, and present ablations over segment length and bootstrap strategy.

Recently, Muon has gained substantial attention as an appealing alternative to Adam-like optimizers, with many works highlighting its advantages through spectral normalization and improved conditioning. Yet this positive theoretical narrative contrasts with its empirical performance in large language model (LLM) training, where Muon's gains over Adam/AdamW are often mixed, schedule-sensitive, and not uniformly superior. To address this gap, we develop a trajectory-level theory characterizing both the strengths and limitations of Muon. We introduce a mixed-spiked matrix sensing model whose sensing operator decomposes into signal, spike, and bulk components, capturing a mixture of anisotropic structure and long-tail information reminiscent of LLM training. On top of it, we adopted a river-valley perspective in which we view the landscape as composed of a river direction flowing to the desired solution and hill directions encoding nuisance or task-irrelevant information. In the momentum-free setting, we show that Muon moves faster along the information-bearing river direction during early optimization, but can converge much more slowly near the river bottom than gradient descent. We then extend the river-valley perspective to general nonconvex objectives with momentum by studying points on the spectral river. There, while Muon converges faster early on, its orthogonalized update removes residual scale information, making it prone to overshooting and oscillation near the target solution. Together, these results suggest that our characterizations extend beyond spiked matrix sensing and motivate switching to GD-like refinement optimizers in the final phase, rather than relying only on a fixed learning-rate schedule for Muon. We also provide preliminary evidence supporting this two-stage approach in language model training experiments.

The rapid proliferation of IoT and IoMT devices introduces critical cybersecurity vulnerabilities in healthcare and industrial environments where resource-constrained devices operate under strict latency and data-privacy regulations. This paper presents the Federated Temporal Graph Convolutional Network with Advantage Actor-Critic (Federated TGCN-A2C), a privacy-preserving defense architecture integrating four mechanisms: a PyG-based Temporal GCN using GCNConv layers with global mean pooling and a learned anomaly gate for flow-level threat classification; LSTM-based Digital Twins generating per-device anomaly scores gating the classifier via learned sigmoid coupling; a Federated A2C agent selecting among ALLOW, ISOLATE, and HONEYPOT-REDIRECT actions based on a seven-dimensional state capturing confidence, entropy, anomaly magnitude, and traffic composition; and an enhanced honeypot layer converting suspicious traffic into threat intelligence with adaptive thresholds. Federated aggregation employs EMA-smoothed per-client validation losses as inverse-weighted FedAvg coefficients to stabilize global model updates under non-IID distributions, with cosine-annealed learning rates per round. Evaluated on CICDDoS 2019 and TON-IoT benchmarks, the framework achieves 99.48% and 99.61% test accuracy with weighted-F1 scores of 0.9948 and 0.9961, converging within 25 and 10 federated rounds, outperforming Fed-Inforce-Fusion by 0.21 percentage points while covering three additional attack categories. All sixteen CICDDoS 2019 classes achieve F1 of at least 0.9237 and all ten TON-IoT classes achieve F1 of at least 0.9488, including the severely imbalanced MITM category. Post-hoc explainability via SHAP, LIME, Grad-CAM, and counterfactual analysis confirms decisions are grounded in semantically meaningful flow features, supporting regulatory accountability in clinical deployments.

Autoregressive text-to-image (T2I) generation has recently advanced rapidly, yet aligning generated images with human preferences remains challenging. GRPO-style online reinforcement learning provides an effective framework; however, existing methods typically treat reference-policy divergence as fixed, despite its direct impact on policy optimization. We study this overlooked factor within a unified f-divergence framework, encompassing forward KL, reverse KL, and JS divergence, for GRPO-style autoregressive T2I alignment. Our systematic theoretical analysis reveals that different divergences reshape token-level updates in distinct ways. In particular, under the sampled-token shaping form used, JS regularization achieves a favorable trade-off by mitigating uniform bias relative to the reference policy while still discouraging large deviations. Extensive experiments on LlamaGen and Janus-7B show that JS divergence achieves the strongest or highly competitive optimization performance on most evaluation metrics while maintaining favorable generation diversity. The code is available at https://github.com/tuoyou-hao/BPD-GRPO.

Modern deep neural networks are trained using error backpropagation, which requires sequential forward and backward computations across network layers. As these networks become deeper, this introduces limitations, since layer-wise updates are strictly interdependent and cannot proceed in parallel. These constraints restrict training procedures to data-parallel schemes, thereby prohibiting model-parallel training. We propose TreeProp, an architecture-agnostic variational learning framework that organizes network layers into a tree-structured hierarchy. During training, TreeProp replaces sequential forward computations and backward gradient propagation with hierarchical computations. This allows intermediate representations and learning signals to be constructed in time complexity of $\mathcal{O}(\log N)$ for a network of $N$ layers. To the best of our knowledge, TreeProp is the first learning algorithm for deep neural networks with logarithmic parallel time complexity for both forward computation and backward gradient propagation during training. Furthermore, we show that multiple valid paths through the hierarchy exist, such that TreeProp implicitly learns subnetworks with different effective depths, but without additional training effort. We evaluate TreeProp on vision classification and autoregressive language modeling, matching the performance of conventional end-to-end training for a variety of tasks and outperforming previous contrastive training approaches. We further demonstrate the applicability of TreeProp to recurrent neural networks that otherwise rely on backpropagation through time.

Deploying generalist robotic agents in the real world requires transferable skills. Specifically, a policy trained to clone a behavior from object-specific demonstrations must generalize beyond that object, otherwise data collection requirements become intractable. Recently, fine-tuning of pre-trained billion-parameter Vision-Language Models (VLMs), initially on large-scale robot datasets and then on fewer scenario-specific demonstrations, has emerged as the predominant paradigm for designing Vision-Language-Action (VLA) models. While these policies achieve state-of-the-art manipulation performance in-distribution, they remain brittle to minor spatial, semantic, and task variations. In this work, we address the inability of current models to decouple the declarative (i.e., concepts and entity semantics) from the procedural knowledge (i.e., how to do something) encoded in their parameters, which is a fundamental bottleneck for zero-shot skill transfer to novel objects. To address this, we propose w$^{2}$VLA, a new VLA model with restructured information flow. Rather than feeding all multimodal tokens from the VLM encoder into a large, opaque transformer-based action expert, our approach modulates the robot state sequence with visual, spatial, and skill information in a compositional and interpretable manner. Unlike popular, state-of-the-art VLAs, we show that our modular approach successfully decouples knowledge representations, enabling robust behavior cloning and unprecedented zero-shot skill transfer capabilities across dissimilar, unseen objects.

Risk stratification for advanced colorectal polyps typically relies on colonoscopy and/or pathology findings. However, there is growing interest in whether non-invasive features available prior to colonoscopy can help identify patients at higher risk. Such approaches may enhance clinical decision-making by prioritizing surveillance for individuals most likely to harbor high-risk polyps, when colonoscopy resources are limited while potentially reducing unnecessary procedures in lower-risk patients. Importantly, the use of non-invasive, pre-procedural information may also help promote more equitable access to risk stratification, particularly in settings where colonoscopy resources are limited or unevenly distributed. We aimed to develop and externally validate machine learning models to predict high-risk colorectal polyps using only non-invasive, pre-colonoscopy demographic, clinical, and behavioral features in a diverse, predominantly African American, urban cohort. We conducted a retrospective cohort study using demographic, lifestyle, and comorbidity data from patients who underwent colonoscopy at Howard University Hospital to develop and validate several machine learning models, including neural networks, random forest, support vector machines (SVM), Naive Bayes, logistic regression, decision trees, k-nearest neighbors (KNN), and XGBoost, for predicting high-risk colorectal polyps. High-risk polyps (HRP) were defined as villous or tubullovillous adenomas, high-grade dysplasia, polyps >= 10 mm in size, and/or the presence of >= 3 polyps per procedure; all other cases were classified as low-risk polyps (LRP). The dataset included 4,681 patients from 2015-2022 used for internal validation and 1,562 patients from 2023-2024 used for external validation.

The vulnerability of ML models to adversarial examples has recently emerged as a major concern. While adversarial training is one of the most effective countermeasures to this issue, its high computational cost remains an obstacle to practical deployment. Recent progress in reducing this cost has relied, in the case of linear models, on a formal equivalence between the adversarial risk and a simpler form of regularized risk. This enabled significantly more efficient training procedures, which naturally raises the question of whether such an equivalence can be extended beyond linear models. In this work, we formally show that no such equivalence is possible for two-layer networks. Our proofs proceed via a reduction to key properties that fundamentally separate the adversarial risk from any simple regularized risk which would only exhibit a weak form of data dependence. Beyond this setting, we provide empirical evidence on Wide-ResNets indicating that the same type of impossibility persists in deeper and more expressive architectures.

Accurate soil moisture estimation in semi-arid agricultural regions requires integrating remote sensing and meteorological information while accounting for the delayed response of soil moisture to atmospheric forcing. This study introduces a Cross-Correlation Function (CCF) methodology to determine optimal temporal lags (0-30 days) between meteorological variables and soil moisture, as well as inter-depth lags (0-15 days) describing vertical moisture propagation from the surface (10 cm) to deeper layers (20-50 cm). The approach was validated across seven agricultural plots in southeastern Spain. Three deep learning architectures, each targeting a distinct prediction granularity, were evaluated under five feature configurations ranging from satellite-only to full satellite-meteorology-depth fusion: a CNN for per-pixel estimation within each plot, an LSTM for frame-level (daily plot-mean) prediction, and a CNN-LSTM hybrid operating on sliding windows with pooled multi-patch training. Models were assessed on held-out data to measure genuine generalisation. Meteorological variables improved performance over the satellite-only baseline, while subsurface depth information proved decisive across all architectures. The per-pixel CNN achieved the strongest single-patch result (R^2 = 0.877, RMSE = 2.28), with a seven-patch average R^2 of 0.535, representing an improvement of +1.00 over the satellite-only baseline. The pooled CNN-LSTM hybrid obtained the highest overall performance (R^2 = 0.930, CVRMSE = 8.0%). These results demonstrate that explicitly modelling atmospheric and vertical subsurface delays substantially improves soil moisture estimation for precision agriculture.

Vision–Language–Action (VLA) controllers are often built by extending vision–language models (VLMs) with action supervision, relying on multimodal backbones with large data and compute requirements. We demonstrate that a text-only large language model (LLM) can be adapted into a VLA-style controller when visual observations are rendered into a text input using an ASCII representation. This ASCII-as-vision interface enables existing training and deployment stacks for LLMs to efficiently condition on visual state, follow natural-language instructions, and produce constrained, executable actions. We fine-tune and compare multiple LLMs and VLMs across model families and scales, using both expert demonstrations from a planning-based teacher, as well as DAgger for iterative improvement. In a 2D manipulation benchmark, in both simulation and on a physical manipulator, the resulting controllers can identify task-relevant entities and plan feasible action sequences. Our results suggest that ASCII rendering can serve as a lightweight, interpretable modality bridge from images to text, complementing conventional VLA pipelines, and opening directions for VLA research with text-only backbones.

Accurate segmentation of subcortical regions is critical for neurosurgical planning and functional research. Most automated methods rely on template space coregistration, which may compromise patient-specific accuracy, particularly in small structures. We identify a need to evaluate whether native space approaches offer a measurable advantage, which we evaluate in the context of movement disorders. We developed two UNet-based segmentation pipelines of the Subthalamic Nucleus (STN) - a common surgical target in Parkinson's Disease - and the neighbouring Red Nucleus (RN) and Substantia Nigra (SN). We collected 7T and 3T MRI data from five public datasets. The pipelines were evaluated in the native-space against manual labels. We further investigated the effect of the template resolution. Motivated by the hypothesis that models may better learn target boundaries in higher field, we tested the transferability of 7T-trained models to 3T clinical images, and whether synthetic 3T training data - generated via a disentangled representation learning method - could help bridging this domain gap. On held-out 7T data, the native pipeline consistently outperformed the template one. For the STN, native-space Dice reached 0.775 +- 0.055 versus 0.713 +- 0.051 (1 mm template), with HD95 of 0.79 +- 0.24 mm versus 1.17 +- 1.10 mm, respectively. Similar advantages were observed for the RN and SN. Increasing template resolution did not improve accuracy. When applied to 3T images, all models showed a considerable performance drop. Adding synthetic 3T data yielded only modest improvements, though without degrading 7T performance. Native-space segmentation is preferable for applications requiring patient specific anatomical fidelity, such as the surgical planning in PD. Bridging the 7T-to-3T domain gap remains an open challenge, motivating future work on domain adaptation tailored to subcortical structures.

Speech enhancement language models achieve strong results when trained on discrete audio tokens, but their optimization relies on token-level cross-entropy rather than the perceptual metrics used for evaluation. We introduce a post-training stage for autoregressive speech enhancement language models using Group Sequence Policy Optimization (GSPO) with multi-metric perceptual rewards. Our method directly optimizes non-differentiable quality metrics (DNSMOS, WER, and UTMOS) as reward signals, without learned surrogates or offline preference pairs. Applied to two autoregressive base models, UniSE and GenSE, our approach achieves state-of-the-art results on the DNS2020 benchmark. A human evaluation ablation further shows that the composite multi-metric reward is preferred over any single-metric variant, confirming that multi-reward optimization avoids the reward hacking observed with single-metric training.

As large language models scale, memory bandwidth for key-value caches and retrieval-augmented generation systems becomes a critical bottleneck. While 1-bit quantization addresses this constraint, recent TurboQuant relies on dense random rotation matrices to condition the vector distribution before quantization. This projection demands millions of floating-point multiplications per embedding, making it difficult to deploy on constrained edge silicon. We introduce Fast-TurboQuant, a multiplier-free projection architecture that replaces the dense matrix with a structured fast Johnson-Lindenstrauss transform. By applying a Rademacher phase inversion followed by a fast Walsh-Hadamard transform (FWHT), the method leverages sub-Gaussian concentration to satisfy the prerequisites of scalar Lloyd-Max quantization without Gaussian projections. This substitution reduces the arithmetic complexity to only additions, eliminating hardware multipliers. Evaluation on DBpedia OpenAI-3 Large embeddings demonstrates a 19.7 times algorithmic speedup under sequential execution. Furthermore, the dimension expansion due to the FWHT zero-padding reduces the mean squared error and improves Recall@10.

Large-scale optimization problems are often solved repeatedly under similar structural conditions, leading to substantial computational overhead. This occurs in applications such as power systems, transportation, and supply chain networks, where the underlying structure is fixed while parameters frequently vary under perturbations. This paper proposes a Learning to Optimize (LTO) framework that accelerates the solution of large-scale general mixed-integer problems by leveraging the concept of a backdoor, i.e., a subset of variables that drive most of the computational complexity. The proposed BIPC framework consists of three phases. Phase I is an identification procedure that discovers a backdoor for a set of instances in the distribution. Phase II uses supervised learning to develop machine learning models that, given an instance, predict values for bounded-domain backdoor variables and intervals for wide-domain backdoor variables. These predictions define a reduced optimization problem where the predictions constrain the backdoor variables, while the other variables remain free. Phase III optimizes this reduced problem and, if necessary, applies a correction step to restore feasibility or the optimality guarantees. Experiments on real-world, large-scale problems show substantial reductions in solution time with only a limited loss in solution quality. The framework enables organizations to solve large-scale optimization problems efficiently in the presence of frequent perturbations, such as unexpected events, demand fluctuations, or operational changes. Because these changes affect parameters rather than the problem structure, BIPC can quickly provide high-quality, feasible solutions, offering a practical approach to integrating machine learning into existing optimization pipelines.

Relational Deep Learning (RDL) models multi-tabular databases as temporal heterogeneous graphs for end-to-end representation learning. While RDL is evolving rapidly, existing approaches face significant generalization obstacles. They are either schema-specific, requiring training from scratch for every new database, or they rely on monolithic architectures that entangle feature encoding with graph message-passing. Analyzing these limitations, we establish four core pillars for building foundational relational models: semantic granularity, structural topology, temporal causality, and unified optimization. Addressing these pillars, we propose a modular approach that decouples row encoding from graph message-passing. We introduce the Universal Row Encoder, a transformer-based module that integrates raw cell data with schema metadata$-$including column semantics, table names, and global distribution statistics$-$to produce table-width invariant row embeddings. By explicitly feeding global statistics to an intra-row self-attention mechanism, the encoder natively contextualizes unseen features and handles sparse data. Serving as a flexible "backend" for any downstream graph architecture, our pretrained encoder enhances cross-database knowledge transfer on the established RelBench benchmarks while improving learning convergence and memory footprint.

In this article, we analyse convergence of the averaged Adam optimizer to an attracting zero of the Adam vector field. We provide a central limit theorem that, in particular, quantifies exactly the speed of convergence. The order of convergence is $n^{-1/2}$ in the number of steps of the algorithm which coincides with the order observed for classical stochastic approximation algorithms. The covariance in the central limit theorem is given in terms of properties of the Adam algorithm in the state of the attractor.

Tensor factorization (TF) has been widely adopted for high-dimensional sparse data completion tasks. Despite significant progress, neural TF methods often struggle to capture complex cross-mode interactions and remain vulnerable to (extreme) data sparsity. To address these challenges, we propose a novel neural tensor factorization approach, termed Dual-Attention Convolution Expert Networks with Group-Level Contrastive Learning (DCGC). For the first problem, DCGC generates diverse non-linear alignment patterns of latent factors via a multi-channel convolution network, and leverages the gated dual-attention mechanism to drive the model to focus on more important output channels (i.e., convolution experts) and the aligned features. Furthermore, DCGC introduces a group-level contrastive learning strategy that aggregates positive samples with identical feedback levels while separating negative samples across different levels. This strategy injects high-quality self-supervised signals to mitigate data sparsity. Extensive experiments conducted on five datasets demonstrate that our DCGC outperforms the state-of-the-art methods in sparse tensor completion for traffic and recommendation applications. Code to reproduce the experimental results in the paper is available at https://github.com/ku1z/DCGC.

The rapid proliferation of Internet of Medical Things (IoMT) devices introduces critical cybersecurity vulnerabilities in healthcare environments where resource-constrained medical devices operate under strict latency requirements and stringent data-privacy regulations. To address these challenges, this paper presents the Lightweight Digital Twin and Federated Reinforcement Learning (LDT-FRL) framework, a privacy-preserving defense architecture integrating four complementary mechanisms: a Temporal Attention Encoder (TAE) built on a GRU backbone with learned temporal self-attention for flow-level threat classification; lightweight LSTM-based Digital Twins trained on normal-class traffic to generate per-device anomaly scores that gate the TAE classifier through a learned sigmoid coupling; a Federated Proximal Policy Optimization (PPO) agent selecting among ALLOW, ISOLATE, and HONEYPOT_REDIRECT actions based on a seven-dimensional state; and an intelligent honeypot layer that converts redirected suspicious traffic into actionable threat intelligence. A federated aggregation strategy employing EMA-smoothed per-client validation losses as inverse-weighted FedAvg coefficients stabilizes global model updates under non-IID client distributions. Evaluated on CICDDoS 2019 and TON-IoT benchmarks, LDT-FRL achieves 99.66% and 99.95% test accuracy respectively, with macro-F1 scores of 0.9913 and 0.9995, converging 81% faster than the DTFL-CD baseline while attaining perfect F1=1.000 on the severely imbalanced MITM class. Explainability analysis via SHAP, LIME, Grad-CAM, and counterfactual methods confirms that the TAE focuses on semantically meaningful flow features, providing interpretable evidence for each defense decision.

Domain shift remains a major obstacle to the reliable deployment of machine learning models in high-stakes environments such as healthcare. While Domain adaptation aims to mitigate these effects, existing approaches suffer from limited expressiveness of latent representations and a reliance on handcrafted, static augmentations. In this work, we address these limitations by proposing a novel deep learning architecture for Unsupervised Domain Adaptation (UDA), specifically optimized for medical image segmentation. Our framework, ADualVUOT, integrates a dual-encoder Variational Autoencoder (VAE) with Continuous Normalizing Flows (CNFs) to increase modeling flexibility and posterior expressiveness. To achieve domain alignment, we leverage Unbalanced Optimal Transport (UOT) through the Gaussian-Gromov-Wasserstein (GGW) distance, which handles structural and topological discrepancies between domains. Furthermore, we incorporate an adversarial augmentation scheme to synthesize worst-case compositions, thus enhancing model robustness. Extensive experiments on medical imaging benchmarks show significant gains over prior OT-based approaches.

Nowadays, large multilingual translation models demonstrate impressive translation capabilities in the machine translation benchmarks. This raises a practical question to the developers: is it worth developing translation models specialized for a particular language pair if you only need to support that language pair? To give an anecdotal answer to this question, we develop a family of small language models (0.8B, 1.4B, 3.3B, and 7B parameters) specialized for Japanese-English bidirectional translation. We employ a two-stage supervised fine-tuning approach followed by Multi-Objective GRPO (Ichihara et al. 2025) to train models on synthetically generated parallel corpora. We evaluate our models on WMT and real-world translation benchmarks across business, legal, medical, financial, and patent domains. While multilingual models achieve strong performance on WMT benchmarks, our compact models outperform them on real-world benchmarks, suggesting the practical utility of developing specialized translation models even in the era of large multilingual models.