Machine Learning
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
Tuesday 11 August
Mixture-of-experts (MoE) models have recently moved beyond routing a fixed number of complete experts. Shared-expert designs preserve reusable knowledge, fine-grained methods vary computation within experts, and dynamic routers adapt the number of active experts. Yet these decisions are usually made independently, overlooking a basic dependency: extracting reusable computation changes both what remains and how much expert capacity the remainder needs. We study this dependency by decomposing sparsely upcycled feed-forward experts into key-value channels. Co-activated experts align at a subset of value positions; removing these positions changes expert preference; and greater shared coverage is associated with lower residual expert demand. These observations lead to one principle: share first, then route what remains. We instantiate it in UniF-MoE, a unified framework for token-adaptive MoE computation. Each expert is partitioned into aligned blocks. A shared-demand score sets the shared block count and pathway weight, key prototypes select the shared content, and the complementary demand determines the residual expert count through cumulative routing mass. A Gram regularizer separates and normalizes router embeddings, promoting diverse routing directions, sparse expert overlap, and a simple routing geometry. Experiments on DomainBed and GLUE show that this unified design improves predictive performance over representative static and dynamic MoEs while reducing activated computation, inference latency, and memory. Code is available at https://github.com/existence0420/UniF-MoE.
Networking
Spatio-Temporal Scheduling for Robust and Efficient Multi-Transmitter Wireless Power Transfer
In multi-user wireless power transfer (WPT), scheduling schemes that determine the allocation of transmission resources among receivers play a crucial role in improving power transfer efficiency. Scheduling can be classified into time-division (TD) and space-division (SD) schemes, with the transmission order and direction designed to control when and to whom power is delivered. TD-WPT can exploit the nonlinear characteristics of rectennas by concentrating power in time; however, a system relying on highly directional transmission from a single location reduces robustness under time-varying channel conditions. This study investigated the effectiveness of spatio-temporal scheduling in coordinated multi-transmitter WPT systems. By employing multiple transmitters, the proposed method is robust against channel variations in delivering power. Moreover, coordinated beamforming among transmitters exploits inter-cluster interference. Simulation results demonstrate the potential of the proposed scheme for robust and efficient power supply, even under shadowing conditions.
Read the full abstract
Numerical Analysis
A fifth-order divergence-free finite difference Hermite WENO scheme for ideal magnetohydrodynamics
In this paper, we present a fifth-order finite difference divergence-free Hermite weighted essentially non-oscillatory (HWENO) scheme for the ideal magnetohydrodynamics (MHD) equations. In this framework, both the solution and its spatial partial derivatives are evolved in time and jointly employed in the spatial reconstruction procedure. A major challenge in MHD simulations is preserving the divergence-free constraint of the magnetic field, which is generally violated by standard numerical methods designed solely for hyperbolic conservation laws. To address this issue, we first solve the MHD equations within the HWENO framework for hyperbolic conservation laws, yielding a magnetic field divergence that remains zero up to high-order accuracy in smooth regions. Subsequently, we apply a correction that evenly distributes the divergence error among the partial derivatives involved in the divergence-free constraint, thereby rendering the magnetic field discretely divergence-free at the new time level. This approach offers several advantages. First, the scheme retains the conservation property, as only the partial derivatives of the numerical solution are corrected, leaving the conserved variables unchanged. Second, the correction applied to the partial derivatives of the magnetic field components introduces only a high-order perturbation, thereby preserving the overall accuracy. Third, the divergence-free treatment significantly enhances robustness, as most benchmark test cases cannot be run stably without such a correction. Fourth, the correction is a simple linear operation applied at each stage of the time integration, incurring negligible additional computational cost. Extensive numerical experiments demonstrate the accuracy, resolution, efficiency, effectiveness, and robustness of the proposed scheme.
Read the full abstract
Numerical Analysis
Efficient Weak-Entropy PINN for Solving Hyperbolic Conservation Laws
In recent years, neural networks have significantly advanced numerical solutions of partial differential equations (PDEs). However, solving PDEs with discontinuous solutions, such as hyperbolic conservation laws, remains challenging for neural network-based methods such as physics-informed neural networks (PINNs). Existing methods often rely on strong prior assumptions such as knowledge of discontinuity locations, or they introduce artificial smoothing terms that degrade accuracy. However, accurately solving these conservation laws and predicting the formation and propagation of discontinuities in solutions is crucial in many practical applications, including gas dynamics and traffic flow modeling. In this paper, we introduce a novel Weak-Entropy PINN (WEPINN) framework for hyperbolic conservation laws with discontinuous solutions. The method enforces the governing equations in their weak (integral) formulation and incorporates the entropy condition to select the physically admissible solution, while employing the discrete fast Fourier transform (DFFT) for efficient numerical integration. Our method is tested through extensive numerical experiments on a variety of scalar conservation laws and systems of conservation laws in one and two dimensional spaces. These experiments demonstrate that our method can accurately resolve sharp discontinuities while effectively capturing interactions between multiple shock and rarefaction waves.
Read the full abstract
Robotics
Hip Energized Monopedal Hopping
We present a novel stepping strategy for pitch unlocked planar monopeds where the reaction torques from stabilizing pitch with a conventional PD + feedfoward controller are recruited to counteract energetic losses from damping. By moving the location of the mass center, our controller increases the pitch stabilization torque, thereby adding energy to the gait. A new stepping policy adjusts the distribution of energy between the radial and angular degrees of freedom to counteract dissipative losses and achieve a user specified balance between steady state fore-aft speed and apex height. Hybrid averaging analysis yields closed form expressions for the fixed points and eigenvalues of the resulting gait, lending insight into the interplay between the physical and control parameters' influence on performance. Simulation studies on a generic 5 link biped and a careful model of the Penn Jerboa reveal a useful correspondence to these analytical predictions. Physical experiments on the Penn Jerboa exhibit stable locomotion with speeds ranging from 1.02 m/s to 1.77 m/s (5.10 leg lengths/s to 8.85 leg lengths/s) in a manner effectively approximated by the mathematical analysis.
Read the full abstract
Machine Learning
Dreamer-SAC: Off-Policy Learning in Latent World Models for Sample-Efficient Autonomous Driving
Sample-efficient reinforcement learning for autonomous driving is often limited by the trade-off between data efficiency and model bias. While world models reduce the reliance on costly environment interactions, policy optimization over learned dynamics remains sensitive to prediction errors. This paper proposes the Dreamer-SAC framework, which integrates a recurrent state-space world model with an off-policy soft actor-critic algorithm trained directly in latent space. The framework uses a combination of real interactions and short-horizon generated trajectories with n-step target estimation and multi-objective supervision. Evaluated in autonomous driving scenarios with objectives encompassing driving efficiency and safety, the proposed framework consistently outperforms representative reinforcement learning baselines, including DreamerV3, SAC, and PPO, while achieving improved performance with substantially fewer real environment interactions. Experiments reveal an inverted-U relationship between rollout horizon and policy performance, where short-horizon latent rollouts achieve the best trade-off between additional training signals and accumulated model bias. Furthermore, n-step target estimation demonstrates more effectiveness over one-step temporal-difference targets in exploiting predicted experience for value learning.
Read the full abstract
Information Retrieval
Persona Conditioning as an Assessor-Sensitivity Probe for LLM-Based IR Evaluation
Large language models (LLMs) are increasingly used as relevance assessors in information retrieval (IR) evaluation, raising questions about how assessor framing affects judgment reliability and downstream system comparison. We study persona conditioning as a diagnostic mechanism for exposing LLM assessor sensitivity. Using task-oriented personas drawn from two complementary sources (PersonaHub and NVIDIA Nemotron-Personas-USA), we instantiate five assessor roles emphasizing intent interpretation, domain expertise, contrastive judgment, evidence verification, and global search-quality assessment, compared with a standard UMBRELA baseline. Across six LLM backbones on TREC DL20 and RAG24, our analyses reveal structured rather than uniform assessor sensitivity. Judgments usually remain close to the baseline while shifting assessment strictness, evidential threshold, or interpretation emphasis rather than producing widespread relevance reversals. At the system level, high-capacity models preserve system-ranking agreement, while smaller models amplify persona-induced instability. Local rank-displacement analysis shows sensitivity concentrates on particular retrieval systems and system types, especially neural ranking/reranking systems on DL20 and RAG-oriented pipelines on RAG24. Persona source matters less than assessor role and model capacity. These findings position persona-conditioned judging as a controlled sensitivity probe for stress-testing LLM-based IR evaluation pipelines and identifying systems whose evaluation outcomes are sensitive to assessor framing.
Read the full abstract
Machine Learning
Generator-Guided Inverse Sampling for Lévy-Driven Generative Models
This paper studies inverse sampling for Lévy-driven generative models from the perspective of Markov generators. Unlike conventional diffusion models, Lévy-driven dynamics involve infinite jump activities, which makes their reverse process nonlocal and difficult to characterize using score information alone. We address this challenge by analyzing the forward and reversed generators. It is derived that the reversed jump component generally becomes a state-dependent Markov jump process governed by a nonlocal density ratio. This observation motivates a structured reverse sampler that decomposes the dynamics into diffusion, small jump, and large jump components. Based on this characterization, we develop a computationally tractable sampler for a class of isotropic linear Lévy SDEs with symmetric $\alpha$-stable jump components. For the jump component, the neural network is used only to amortize the rate of large jump activities, while jump amplitudes are generated from analytically derived conditional distributions, which improves interpretability and controllability. Efficient implementation techniques are further introduced under this setting to avoid expensive high-dimensional integration and sampling. The sampler is further adapted to approximate observation-guided sampling and applied to OFDM-SISO channel estimation under mixed Gaussian and impulsive noise. Simulations show robust estimation performance with a favorable tradeoff between complexity and performance.
Read the full abstract
Robotics
Learning Cooperative Bimanual Dexterous Grasp of Large Objects with Tactile Sensing
Bimanual dexterous grasping of large objects is a critical challenge in robotic manipulation. However, most existing studies focus on sequential manipulation rather than cooperative grasping, and methods addressing such bimanual tasks have largely been limited to simulation. These limitations stem from the difficulty of acquiring full 3D object models and generating physically plausible grasping actions. To fill this gap, we propose a real-world bimanual grasping framework that includes: a multimodal dataset capturing joint angles, visual observations and force signals; a Denoising Diffusion Probabilistic Model (DDPM)-based module that generates joint-level grasp configurations from segmented point clouds; and an execution strategy that integrates motion planning with online grasp refinement to ensure physical stability and feasibility. Our approach enables the synthesis of executable bimanual grasps from single-view inputs, reducing dependence on complete 3D object models and ensuring stable real-world performance. Experiments on a dual-arm robot demonstrate high success rates across unseen objects with varying geometries and poses, and ablation studies confirm the contributions of key components of our system.
Read the full abstract
Artificial Intelligence · Cross-listed from quant-ph
A Single Atom in Front of a Mirror is a Universal Reservoir Computer
Universal approximation in reservoir computing is typically associated with a class of reservoirs. We show that universality can be associated with a single reservoir, considering a minimal setup of a single atom in front of a mirror. In its linear-transducer limit, our reservoir is a universal approximator of fading-memory maps under an operating class of checkable conditions, with a rate constant measured at the operating point. A given reservoir can reach arbitrary accuracy by changing measurement settings. The proof gives an explicit recipe: for a target accuracy, it specifies the required physical resources and resonator modes. Enlarging the number of accessible modes increases the matchable kernel span without reducing capability. Beyond the linear limit, the atom's saturation replaces high-order polynomial readouts, and the device operates on real-world tasks alongside classical baselines. Our results highlight an example of universality with a minimal quantum setup.
Read the full abstract
Discrete Mathematics
Multiway Partition is FPT for Connectivity Functions
Let $E$ be a set of $n$ elements, $f\colon2^E\to\mathbb Z_{\ge0}$ be submodular, $t_1,\ldots,t_r\in E$ be distinct, and $k\in\mathbb N$. The Submodular Multiway Partition problem asks whether $E$ has a partition $(P_1,\ldots,P_r)$ with $t_i\in P_i$ for all $i$ and $\sum_i f(P_i)\le k$. We prove that this problem can be solved in $r^{O(k)}n^{O(1)}$ time whenever $f$ is posimodular. This covers monotone submodular functions, matroid rank functions, and nonnegative combinations of monotone and symmetric submodular functions. When $f$ is a connectivity function, we remove the dependence on the number of terminals and obtain a $k^{O(k)}n^{O(1)}$-time algorithm. The proof is entirely elementary, and is arguably the simplest proof that Edge Multiway Cut is fixed-parameter tractable. As a new consequence, our main result immediately implies that Hypergraph Multiway Partition is fixed-parameter tractable.
Read the full abstract
Formal Languages and Automata · Cross-listed from math.GR
Conjugacy languages in free inverse monoids
We initiate the study of conjugacy languages in free inverse monoids. Motivated by the notion of conjugacy languages in groups and the study of conjugacy in semigroups, we introduce the language of shortest representatives of conjugacy classes and study it for free inverse monoids of rank at least $2$ under the natural notion of conjugacy. We show that, contrary to the free group case, where this language is regular, in a free inverse monoid it is neither context-free nor co-context-free. In the monogenic case, this language is context-free. We show that this non-context-freeness comes from elements with nontrivial conjugacy classes. We define an equivalence relation as follows: all elements with a nontrivial conjugacy class are related and elements with trivial conjugacy class are only related to themselves. We call this relation $\mathrm{UConj}$. We show that the language consisting of geodesics representing elements whose conjugacy class is trivial is context-free by providing an explicit context-free grammar generating it. For groups, we show that the language of minimal representatives of $\mathrm{UConj}$ classes is regular if the group is hyperbolic and for right-angled Artin groups with the standard generating set, it is piecewise testable. For virtually abelian groups, we show that there is a generating set for which this language is piecewise excluding and exhibit an example of a virtually abelian group admitting a generating set for which this language is not regular.
Read the full abstract
Data Structures and Algorithms
Online Interval Selection on a Simple Chain
A set of intervals $I = \{ I_1, I_2, \dots, I_n \}$ forms a simple chain if, for every $2\leq i \leq n-1$, interval $I_i$ overlaps only with $I_{i-1}$ and $I_{i+1}$. We show that a deterministic memoryless one-directional revoking algorithm achieves a competitive ratio of $2(1 - 1/\sqrt{e}) \approx 0.786$ on the simple chain in the random order model, hence performs worse than the basic greedy algorithm without revoking that has a competitive ratio of $(1 - 1/e^2) \approx 0.864$, but better than any deterministic revoking algorithm in the adversarial model that has a competitive ratio of at most $0.75$. The proof of the latter also leads to a lower bound of $n/4$ for the advice complexity.
Read the full abstract
Computational Engineering
Beyond Forecasting: Recasting Volatility Control as a Routing Problem
Volatility control converts risk estimates into portfolio exposure, yet existing approaches often rely on a fixed volatility estimator or a pre-defined control rule that may not adapt to changing market conditions. We propose VolRouter, a modular framework that formulates volatility control as state-conditioned routing over estimator-controller pairs. VolRouter first summarizes market conditions into a control-relevant state profile and then performs routing through three stages: state inference, switch review, and pair selection. The Router can be implemented using rule-based, learnable, or LLM-based decision modules, while portfolio actions remain generated by predefined control policies. We evaluate VolRouter across S&P 500, Multi-Asset, Bitcoin, and USDT volatility-control settings. VolRouter achieves the highest Sharpe ratio in three of four settings. On S&P 500, it improves Sharpe from 0.952 for RV + Naive Scaling to 1.222 while reducing maximum drawdown from 15.10% to 12.58% and daily CVaR from 1.76% to 1.32%. On Multi-Asset, it improves Sharpe from 1.498 to 1.540 and reduces CVaR from 1.56% to 1.18%. Bitcoin shows similar improvements in risk-adjusted performance, while USDT provides a boundary case where simpler state-aware selectors remain competitive. Ablation and sensitivity analyses show that the improvement comes from relative policy evaluation and selective persistent switching rather than simply expanding the policy library. These results suggest that volatility control can be viewed as a policy-selection problem when risk management requirements vary across market states.
Read the full abstract
Machine Learning
Fisher8: Stabilizing Neural Heteroscedastic Regression via Output-Layer Fisher Geometry
Training neural networks to jointly predict mean and uncertainty estimates from noisy observations can be unstable, prompting a series of independent stabilization efforts. We argue that these interventions highlight a common underlying issue where gradient steps are poorly aligned with the geometry of the loss landscape. To better align updates with local curvature, we derive Fisher8, an output-layer gradient correction that reorients and rescales updates using Fisher geometry rather than Euclidean geometry. Unlike past stabilizers, Fisher8 introduces no data-dependent hyperparameters beyond learning rate and admits an approximate KL trust radius between successive predictive distributions. We show that prior stabilizers converge on overlapping components of this geometric correction. Across multidimensional regression and representation-learning tasks, Fisher8 obtains superior likelihood–error tradeoffs, predicts calibrated uncertainty estimates, and learns rich uncertainty-aware feature spaces.
Read the full abstract
Game Theory
The Game of Marginal Utilities
We study a noncooperative resource-allocation game in which $m$ players distribute fixed resources among $n$ projects and the payoff of player $j$ is given by \[ F^j(x) = \sum_{i=1}^n \frac{a_i x_i^j}{b_i+\sum_{\ell=1}^m x_i^\ell}, \] where $a_i$ and $b_i$ are project parameters, while $x_i^j$ is the amount of resources player $j$ allocates to project $i$. This specification combines diminishing returns with congestion generated by competitors. We prove that the game has a unique Nash equilibrium and characterize it by an equimarginal principle. We show that, after the projects are ordered by $a_i/b_i$, each player invests in an initial segment of projects, and these segments are nested across players, so the equilibrium decomposes into consecutive activity zones; players with larger resources invest weakly more in every project. In the fully active regime, where every player invests in every project, we reduce the equilibrium to a single scalar nonlinear equation for the aggregate marginal-utility rate; all individual marginal rates, investments, and payoffs then follow from explicit formulas. We also provide a projected marginal-utility algorithm with global linear convergence under an explicit step-size condition, together with a structure-exploiting Block Pandora algorithm that reconstructs and certifies the equilibrium conditional on a proposed nested cutoff structure.
Read the full abstract
Machine Learning
Invertible Logits Transformation for Accuracy-Preserving Post-Hoc Uncertainty Calibration
Post-hoc calibration aligns a classifier's predicted confidences with its empirical accuracy without retraining. An ideal calibrator should correct nonlinear miscalibration, scale gracefully to large label spaces, and preserve the original predictions; existing methods typically violate at least one of these properties—temperature scaling lacks expressivity, more flexible parametric alternatives introduce parameters that grow with the number of classes $C$, and other expressive methods do not preserve the rank ordering of class scores and may alter the predicted class. We propose Invertible Logits Transformation (InvLT), which applies a learned scalar MLP $f:\mathbb{R}\to\mathbb{R}$ element-wise to the pre-softmax logits. Sharing $f$ across all logit dimensions makes the parameter count independent of $C$. Monotonicity of $f$—and hence preservation of the argmax prediction—is softly encouraged via a paired inverse network rather than enforced through the numerical integration required by prior monotone calibrators; this avoids their computational overhead while empirically preserving the original classification accuracy in every setting we evaluate. Across standard image classification benchmarks and a range of architectures, InvLT consistently outperforms a broad set of post-hoc baselines on standard calibration metrics.
Read the full abstract
Human-Computer Interaction
Visual-to-Haptic Augmentation in XR: A Wearable Glove for Perceptual Grounding in Multimodal Interaction
Extended Reality (XR) systems increasingly deliver high-fidelity visual and auditory experiences, yet tactile perception remains comparatively underutilized as a modality for enriching embodied interaction. This work presents a visual-to-haptic wearable glove and a feature-based visual-to-haptic mapping algorithm that translates spatial and temporal visual features from images and videos into distributed vibrotactile patterns. The proposed method extracts motion, edge, and brightness cues and fuses them into actuator-level intensity maps aligned with a 29-actuator glove arranged in a five-by-seven layout. The system is implemented through a modular four-layer architecture comprising the XR environment, media content handling, visual-to-haptic processing, and embedded haptic hardware. A within-subject user study (N = 20) compared visual-only interaction with visual-plus-haptic augmentation across texture-based and dynamic video scenarios. Results indicate that tactile augmentation significantly improves perceived realism in dynamic video scenarios and enhances immersion and visual-tactile correspondence across conditions, with stronger and more consistent effects observed for dynamic visual events. While the current implementation operates in a single-user, offline-synchronized configuration, the findings demonstrate that vision-driven tactile augmentation can function as a perceptual enhancement layer within multimodal XR systems. Such a layer may provide a foundation for future socially enriched XR environments where coherent multisensory grounding supports higher-level interaction and communication.
Read the full abstract
Robotics
A Neural Network Based Teleoperation for Remote Controlled Vehicles
Direct teleoperation of vehicles faces critical technical bottlenecks: communication latency and the operator's inability to physically perceive unmodeled environmental disturbances (e.g., aerodynamic drag, bank angles) coupled with highly nonlinear tire-road dynamics. To address these challenges, we propose a tailored unilateral teleoperation framework. The system integrates the Wave Variable (WV) approach to passively guarantee stability under stochastic delays, and an adaptive Radial Basis Function Network (RBFN) to actively compensate for vehicle-specific uncertainties. Unlike existing WV-neural network architectures designed for bilateral robotic arms, our framework features decoupled adaptive laws specifically designed for vehicle longitudinal and lateral dynamics. Furthermore, compared to model-heavy predictive controllers, the model-free RBFN offers rapid online adaptation without heavy computational overhead. Building upon our preliminary theoretical formulation, this brief paper presents comprehensive comparative analyses and real-world hardware validations. Simulation benchmarks against PID, LQR, MPC, and NMPC demonstrate that the RBFN achieves superior robustness against unmodeled disturbances while requiring orders of magnitude less execution time than MPC and NMPC, making it ideal for resource-constrained vehicle edge computing. Finally, hardware-in-the-loop experiments using a 1/10th scale vehicle over a 4G network validate the system's practical feasibility, safety, and robust trajectory tracking under physical road uncertainties.
Read the full abstract
Artificial Intelligence
DSAgentBench: Can Agents Automate End-to-End Data-Science Workflows in Real Computer Environments?
Real-world data science involves long-horizon workflows that span data wrangling, exploration, modeling, visualization, and validation, and require coordinated use of tools such as notebooks, IDEs, terminals, browsers, and databases within real operating environments. Yet existing benchmarks lack real-computer interaction and do not evaluate whether agents can execute complete end-to-end data-science workflows in realistic computing environments, failing to capture the multi-stage, multi-tool nature of data-science practice. We introduce DSAgentBench, the first benchmark to evaluate whether agents can automate full data-science workflows inside real computer environments. DSAgentBench contains 275 diverse tasks covering the entire data-science life-cycle, reflecting the complexity and tool coordination required in practice. Each task requires grounding decisions in intermediate outputs and coordinated tool use, and includes a deterministic evaluator that verifies analytical correctness, visual outputs, and model performance rather than code-only execution. Our extensive experiments with 15 closed- and open-source models show that even the strongest agent, Claude-4.6-Sonnet, achieves only 56.70% task success, while all open-source agents remain below 1%, frequently failing at tool orchestration, OS grounding, and multi-step reasoning. These results reveal a substantial capability gap between current agentic systems and real data-science workflows, positioning DSAgentBench as a foundation for developing grounded, verifiable, autonomous data-science agents. We release DSAgentBench at https://github.com/vis-nlp/DSAgentBench.
Read the full abstract
Artificial Intelligence
Nutrition Data Infrastructure for the AI Era: Operationalizing FAIR for Agent-Mediated Research
AI agents can accelerate nutrition research, but their analyses inherit the identity, semantic, and release ambiguities of the underlying data. We present Nutrition Data Service (NDS), source-preserving infrastructure that operationalizes FAIR for automated use: description resolution makes release-specific records findable; typed crosswalks connect independently released resources; machine-readable interfaces expose versioned sources and crosswalks, supporting replayable and auditable analyses. On food-description benchmarks, NDS outperforms the best published language-model result on NutriBench. External and blinded crosswalk evaluations show that its typed contract favors defensible links and rejects unsupported mappings. In a person-level glycemic-index analysis, pinned NDS inputs produce identical outputs across models and repeated runs, while open-web reconstruction remains unstable. Together, these results show that agent-mediated nutrition research requires a new infrastructure that makes data identity, search, and crosswalk policy explicit.
Read the full abstract
Operating Systems
MemSpec: Memory-Aware Runtime for Adaptive Draft Scheduling in Speculative Decoding on Edge Devices
Speculative decoding accelerates autoregressive large language model (LLM) inference by using a lightweight draft model to speculate multiple tokens, reducing expensive target model decoding steps. Its effectiveness depends heavily on draft selection, motivating adaptive methods that exploit variation across inputs and generation stages. On memory-constrained edge devices, however, these methods often fail to improve end-to-end throughput due to the overhead of switching between draft models. We identify a key limitation in this setting: the mismatch between draft selection and draft availability under tight memory budgets. To address this challenge, we present MemSpec, a prediction-guided, memory-aware runtime for adaptive speculative decoding on edge devices. MemSpec decouples draft selection from execution through proactive resident working-set management. A lightweight predictor estimates draft effectiveness from prompt and generation context, while a memory-aware scheduler reduces reactive model loading overhead. Experiments on a Jetson Orin Nano show that MemSpec improves steady-state generation throughput by 40.7% on average over state-of-the-art bandit-based adaptive methods while closely approaching the oracle upper bound.
Read the full abstract
Human-Computer Interaction · Cross-listed from eess.SP
Neural implants and human safety: single-fault detection for DC-coupled recording front ends
DC-coupled analogue front ends (AFEs) for neural implants provide a low-area solution. However, removing the coupling capacitor eliminates the intrinsic barrier that protects cortical tissue: a single-fault event, such as gate-oxide breakdown of a low-noise amplifier (LNA) input transistor, can open a direct DC path from the supply rail into the brain. On the stimulation side this hazard is well understood, and single-fault tolerance is enforced by a series DC-blocking capacitor; on the recording side, DC-coupled front ends discard the equivalent safeguard, yet their protection has gone almost unexamined. This paper presents a single-fault detection mechanism that monitors the LNA for the DC imbalance produced by such a failure and disables the amplifier before the resulting fault current can irreversibly damage tissue. The imbalance is encoded in the duty cycle of a current-starved relaxation oscillator and read out as a time-to-digital measurement. Designed in 65 nm, the mechanism resolves a worst-case fault of 6.4 nA across all corners within 0.81 ms - compliant with the ISO~14708-3 limit for an 8533 um2 electrode - opening a broader discussion of safety in DC-coupled recording.
Read the full abstract
Human-Computer Interaction
MazzikaAI: A knowledge-based performance-to-prompt compiler for real-time Arabic maqam accompaniment with a streaming text-to-music model
Arabic maqam music microtonal, modal, and built on ornamented call and response is among the traditions most underserved by generative music models, whose training frameworks remain predominantly Western and equaltempered. Real time accompaniment sharpens this gap: an AI partner must listen, adapt dynamically, and respect idiomatic microtonal structures. Streaming text to music models provide strong generative capabilities but lack precise control interfaces. We present MazzikaAI, a knowledge based system that uses natural language as the actuator of a realtime control loop. By compiling live MIDI, gesture, and inferred harmony into continuously updated text prompts, MazzikaAI steers an unmodified streaming generator, Google Lyria RealTime, without requiring model finetuning. The system embeds expert knowledge of six core maqamat, characteristic ornaments, and ensemble dynamics, maintaining realtime responsiveness with subsecond keytoaudibleupdate latency. Empirical evaluations demonstrate that dynamic prompt compilation reliably grounds generation in microtonal scales, significantly increasing offgrid quartertone content over baseline generation. Beyond its core implementation, MazzikaAI illustrates how deterministic knowledgebased rules can effectively bridge expert, nonWestern musical traditions and unfinetuned foundation models. This architecture establishes a scalable paradigm for realtime humanAI cocreation, offering a generalizable blueprint for interactive accompaniment, adaptive music education, and culturally inclusive generative audio across diverse global idioms.
Read the full abstract
Sound
VoxSumm: A Multilingual Corpus of Long-Form Spoken News for Joint Summarization and Translation
As information increasingly traverses linguistic boundaries, users require concise cross-lingual representations of long-form content. Nevertheless, long-document summarization research remains text-centric, whereas multilingual speech research has largely prioritized translation, preserving source content rather than compressing it. We address this methodological gap by formalizing joint speech summarization and translation (JSumT): the generation of a succinct, faithful target-language summary directly from a long spoken document in a source language. We additionally introduce VoxSumm, the first multilingual and cross-lingual benchmark for this task, comprising 10,045 BBC article-summary pairs across 24 languages and encompassing approximately 703 hours of speech data. Our evaluation of representative speech-language models reveals pronounced variation across models and generation settings: Gemini3.1-Pro demonstrates the greatest consistency, summarization into English generally surpasses generation into non-English target languages, and translating an entire document before summarization compounds instruction-following failures. Through the release of VoxSumm, we establish a foundation for developing and evaluating multilingual systems capable of jointly interpreting, compressing, and translating long-form speech.
Read the full abstract
Machine Learning
Efficient Reinforcement Learning for Long-Horizon Tool-Use Agentic Tasks
Long-horizon tool-using agents must reason over user goals, domain policies, tool calls, simulator state, and delayed verifiable rewards. Reinforcement learning (RL) is a natural fit for this setting, but multi-turn on-policy rollouts create long contexts, while model-specific attention layers may require custom masks and learned sink normalization. We present SINKFLEX-RL, a modular training system for RL in dual-control tool-use environments. The system combines a Gymnasium-compatible environment wrapper, a VERL-style rollout dataflow, group-relative policy optimization without a separate value model, and a sink-aware FlexAttention path designed to preserve model-specific sink scaling under causal and sliding-window masks. In a preliminary Tau2Bench retail run, validation reward (mean@1) rises from 0.25 early in training to $0.44$ later in the observed training window, while training-score and trajectory-reward proxies also trend upward. In a fixed-configuration memory benchmark, the optimized attention path reduces peak VRAM from 28.06GB to 22.52GB at 4096 tokens, a $19.7\%$ reduction, and runs the measured 8192-token configuration using $25.53$~GB where the eager baseline runs out of memory. These results illustrate the value of integrating environment interfaces, RL dataflow, and attention-kernel design for memory-feasible long-horizon agent training.
Read the full abstract
Machine Learning
Accelerated Learning of High Dimensional Functions with a Tensor-Featured Training Network
In this work we present a method to accelerate the optimization of learning high dimensional functions using deep neural network (DNN). This optimization procedure introduces contextual features into the first layer of a DNN. The parameters of DNN are optimized via standard gradient descent while keeping the input-feature basis fixed. After optimization of the DNN parameters, the feature layer is provided a chance to update and change before DNN optimization resumes. The feature layer has two types of functions: those that can be evaluated quickly in a matrix-free way on the domain (i.e. rank-1 features) and more complex features that must first be decomposed using tensor network (TN) decomposition strategies (tensor features). In particular, we study the effect of adding features which distill pretrained DNN into TNs using a discretize and decompose strategy. To efficiently decompose high-dimensional functions constructed from discretized DNN, we leverage a randomized tensor decomposition strategy. Using randomization, we are able to reduce the storage cost of decomposing high dimensional functions by at least 8 orders of magnitude. Using this approach, we are able to efficiently train models between 5 and 40 dimensions.
Read the full abstract
Cryptography and Security
Beyond Detection Accuracy: Measuring Explanation Cost, Stability, and Utility for Resource-Aware IoT Intrusion Detection
Machine-learning intrusion-detection studies commonly emphasize predictive accuracy while treating explanation generation as a computationally free post-processing step. This study jointly evaluates predictive effectiveness, explanation cost, local explanation stability, and selective explanation for binary Internet of Things (IoT) intrusion detection. A leakage-safe CICIoT2023 corpus was constructed using exact 39-feature hashes, non-finite-value handling, exact-feature deduplication, conservative label-collision removal, and deterministic hash-level partitioning. Logistic Regression, Decision Tree, Random Forest, and XGBoost were evaluated on natural and balanced test distributions. TreeSHAP cost was measured, stability was assessed under prediction-preserving perturbations, and validation-calibrated policies were used to allocate explanation workload. XGBoost provided the strongest overall predictive profile, while Random Forest produced the lowest false-positive rate. At 5,000 samples, TreeSHAP required 700.759 s for Random Forest and 1.471 s for XGBoost. Random Forest showed the strongest overall base-level explanation stability; XGBoost retained high rank and directional consistency but showed greater top-feature turnover and attribution-magnitude drift. On the balanced test, about 90% false-negative explanation coverage permitted 28-32% compute savings, while about 95% coverage permitted 15-23% savings. Savings were much smaller under the attack-heavy natural prevalence. These results show that operationally useful explainable IoT intrusion detection depends on predictive quality, explanation cost, local stability, workload prevalence, and selective invocation rather than detection accuracy alone.
Read the full abstract
Computer Vision
Towards Unified Dynamic Face Landmark Detection
Although advancements in face landmark detection (FLD) methods continue to push performance boundaries, they overlook two major functional limitations: (1) different network parameters need to be trained independently for each “$N$-point” benchmark dataset, and (2) a model trained on an “$N$-point” dataset reliably outputs only the $N$ landmarks. In our work, we first conceptualize Face Part-Anchored Landmark Positions (FPALPs), wherein each landmark is treated as a progression value between zero (start) and one (end) along a face part's contour. Every landmark can be expressed in the FPALP format, irrespective of its source dataset, hence unlocking the ability to unify all “$N$-point” datasets into a single dataset. Secondly, we represent each landmark with an FPALP-based query, refine it progressively with a cross-modality decoder, and predict its coordinates based on the final representation. Our approach, called Unified Dynamic FLD, embodies these two design choices and streamlines the landmark detection pipeline by enabling (1) a single model to learn on any number of “$N$-point” datasets, and (2) yield any number of specific landmark predictions by loading the designated landmark queries at runtime. Extensive experiments on multiple benchmark datasets show that our method delivers these benefits while remaining competitive with, and in several cases outperforming existing state-of-the-art methods.
Read the full abstract
Computer Vision
CasDeblurGS: Cascaded 2D-to-3D Multi-View Consistency for 3D Gaussian Splatting from Two Blurry Images
Free-viewpoint 3D scene media is increasingly important for immersive applications, yet practical capture often suffers from severe view sparsity and motion blur. Although neural rendering has advanced sparse-view synthesis, existing blur-aware methods typically require substantial multi-view redundancy, accurate camera poses, or costly per-scene optimization. We address a stringent yet practical setting: reconstructing a coherent 3D scene from only two motion-blurred images with known intrinsics, without input-view poses, auxiliary sharp images, or per-scene test-time optimization. To this end, we propose CasDeblurGS, a cascaded framework that progressively recovers reliable cross-view information from local 2D correspondences to global 3D guidance. Stage 1 constructs locally reliable guidance through occlusion-aware correspondence filtering, while Stage 2 aggregates the intermediate restorations into a provisional pose-free 3D Gaussian representation whose input-view re-renders provide dense global guidance for final restoration. The resulting views enable a more coherent 3D representation and higher-quality novel-view synthesis. Experiments on real-world and synthetic Deblur-NeRF scenes show consistent gains over strong baselines, improving PSNR by 1.19 dB and 2.11 dB, respectively. Progressive ablations, cross-view correspondence visualization, and camera reprojection analysis further demonstrate improvements in both rendering quality and multi-view geometric consistency.