arXiv:2509.15330v2 Announce Type: replace Abstract: Recent advances in pre-training vision-language models (VLMs), e.g., contrastive language-image pre-training (CLIP) methods, have shown great potential in learning out-of-distribution (OOD) representations. Despite showing competitive performance, the prompt-based CLIP methods still suffer from: i) inaccurate text descriptions, which leads to degraded accuracy and robustness, and poses a challenge for zero-shot CLIP methods. ii) limited vision-language embedding alignment, which is one important factor affecting generalization performance. To tackle the above issues, this paper proposes a novel Conditional Domain prompt Learning (CoDoL) method, which utilizes readily-available domain information to form prompts and contributes to improved vision-language embedding alignment, which we identify as one factor underlying the observed OOD generalization gains. To capture both instance-specific and domain-specific information, we further propose a lightweight Domain Meta Network (DMN) to generate input-conditional tokens for images in each domain. Extensive experiments on four OOD benchmarks (PACS, VLCS, OfficeHome, and DigitDG) validate the effectiveness of our proposed CoDoL method in terms of empirically improves vision-language embedding alignment across four DG benchmarks, which we present as a contributing factor (rather than the sole cause) of the observed OOD gains.
Science Journals
arXiv:2408.13378v5 Announce Type: replace Abstract: Workflows in drug-target interaction (DTI) assessment require integrating heterogeneous data from predictive models, curated resources, and observations from experimental literature. This evidence can be incomplete or conflicting. DrugAgent is a large language model (LLM)-based multi-agent system focused on DTI evidence integration that integrates outputs from machine learning, knowledge graph, and retrieval-augmented generation (RAG) agents. DrugAgent converts agent outputs into interpretable representations, then summarizes conflict across the evidence. We evaluated DrugAgent on kinase screening data of 900 pairs spanning 178 kinases and 42 inhibitors, and an androgen receptor antagonist screening benchmark. On the kinase dataset, LLM-as-a-Judge evaluation indicated outputs were faithful to input evidence in 98.8% of cases. Biological plausibility of returned summarization was high (scores 3-4 out of 5) across ground-truth classes: 79% of Weak activity labels cases (81% for Moderate/77% Strong); Strong cases received higher scores than Weak/Moderate. Label stability showed 98% agreement across runs. Results on the antagonist benchmark were consistent with the kinase dataset. Retrieved literature provided the greatest benefit when direct drug-target evidence was available, highlighting the importance of evidence availability for RAG-based integration. DrugAgent provides heterogeneous evidence-grounded DTI assessment, complementing standalone DTI prediction. We provide strategies to model agreement, conflict, and uncertainty in biomedical evidence integration. Code: https://github.com/sciluna/DrugAgent.
arXiv:2409.14086v2 Announce Type: replace Abstract: There have been several studies on automatically generating piano covers, and recent advancements in deep learning have enabled the creation of more sophisticated covers. However, existing automatic piano cover models still have room for improvement in terms of expressiveness and fidelity to the original. To address these issues, we propose a learning algorithm called AMT-APC, which leverages the capabilities of automatic music transcription models. By utilizing the strengths of well-established automatic music transcription models, we aim to improve the accuracy of piano cover generation. Our experiments demonstrate that the AMT-APC model reproduces original tracks more accurately than any existing models.
arXiv:2412.13533v4 Announce Type: replace Abstract: Medical image segmentation is a fundamental task in numerous medical engineering applications. Recently, language-guided segmentation has shown promise in medical scenarios where textual clinical reports are readily available as semantic guidance. Clinical reports contain diagnostic information provided by clinicians, which can provide auxiliary textual semantics to guide segmentation. However, existing language-guided segmentation methods neglect the inherent pattern gaps between image and text modalities, resulting in sub-optimal visual-language integration. Contrastive learning is a well-recognized approach to align image-text patterns, but it has not been optimized for bridging the pattern gaps in medical language-guided segmentation that relies primarily on medical image details to characterize the underlying disease/targets. Current contrastive alignment techniques typically align high-level global semantics without involving low-level localized target information, and thus cannot deliver fine-grained textual guidance on crucial image details. In this study, we propose a Target-informed Multi-level Contrastive Alignment framework (TMCA) to bridge image-text pattern gaps for medical language-guided segmentation. TMCA enables target-informed image-text alignments and fine-grained textual guidance by introducing: (i) a target-sensitive semantic distance module that utilizes target information for more granular image-text alignment modeling, (ii) a multi-level contrastive alignment strategy that directs fine-grained textual guidance to multi-scale image details, and (iii) a language-guided target enhancement module that reinforces attention to critical image regions based on the aligned image-text patterns. Extensive experiments on four public benchmark datasets demonstrate that TMCA enabled superior performance over state-of-the-art language-guided medical image segmentation methods.
arXiv:2602.06371v2 Announce Type: replace Abstract: Large Language Models (LLMs) are increasingly deployed in multilingual contexts, yet their consistency across languages on politically sensitive topics remains understudied. This paper presents a systematic bilingual benchmark study examining how 17 LLMs respond to questions concerning the sovereignty of the Republic of China (Taiwan) when queried in Chinese versus English. We discover significant language bias -- the phenomenon where the same model produces substantively different political stances depending on the query language. Our findings reveal that 15 out of 17 tested models exhibit measurable language bias, with Chinese-origin models showing particularly severe issues including complete refusal to answer or explicit propagation of Chinese Communist Party (CCP) narratives. Notably, only GPT-4o Mini achieves a perfect 10/10 score in both languages. We propose novel metrics for quantifying language bias and consistency, including the Language Bias Score (LBS) and Quality-Adjusted Consistency (QAC). Our benchmark and evaluation framework are open-sourced to enable reproducibility and community extension.
arXiv:2602.06872v2 Announce Type: replace Abstract: We derive a residual-based $hp$-a posteriori error estimator for hybrid high-order (HHO) methods on simplicial meshes applied to the biharmonic problem posed on two- and three-dimensional polytopal Lipschitz domains. The a posteriori error estimator hinges on an error decomposition into conforming and nonconforming components. To bound the nonconforming error, we use a $C^1$-partition of unity constructed via Alfeld splittings, combined with local Helmholtz decompositions on vertex stars where the key contribution is to show that the stability constant only depends on the mesh shape-regularity. For the conforming error, we design two residual-based estimators, each associated with a specific interpolation operator. In the first setting, the upper bound on the conforming error involves only the stabilization term and the data oscillation, but hinges on an assumption that we verify numerically. In the second setting, the bound additionally incorporates bulk residuals, normal flux jumps, and tangential jumps. Numerical experiments confirm the theoretical findings on the error upper bound and also illustrate numerically that the proposed estimators lead to moderate effectivity indices.
arXiv:2602.06934v5 Announce Type: replace Abstract: Grassroots Logic Programs (GLP) is a concurrent logic programming language in which logic variables are partitioned into paired readers and writers. An assignment is produced at most once via a writer and consumed at most once via its paired reader, and may contain additional readers and/or writers. This enables the concise expression of rich multidirectional communication modalities. The language was introduced together with concurrent (cGLP) and multiagent (maGLP) operational semantics. Here, we derive from these (\ia)~dGLP, a deterministic counterpart of cGLP, and (\ib)~madGLP, a counterpart of maGLP in which deterministic agents communicate solely by asynchronous message passing, and prove them correct against their abstract counterparts. maGLP shared variable pairs spanning agents can be implemented as local variables paired by \emph{global links}, with correctness following from disjoint substitution commutativity (a consequence of GLP's single-occurrence invariant). We further prove that madGLP is grassroots. Both dGLP and madGLP serve as formal specifications for an AI-driven implementation discipline (math $\to$ informal spec $\to$ Dart) employed and described here: from dGLP, AI (Claude) developed a workstation-based GLP implementation in Dart, and from madGLP it is developing a smartphone-based multiagent one.
arXiv:2607.04969v1 Announce Type: new Abstract: The training paradigm of large language models has shifted from traditional one-pass training to multi-epoch training, as reasonable reuse of limited high-quality data can improve both model performance and sample efficiency. Meanwhile, excessive repetition introduces the risk of overfitting and diminishing returns. Determining when and how to reuse data effectively thus emerges as a natural but under-explored question. Through a novel observation of model's "Memorization Window" signals derived from loss retention dynamics and downstream evaluation scores, we propose "Memorization-guided Data Reuse", a training paradigm that adaptively determines when and how data should be reused, enabling principled decisions on the number of training epochs and the scheduling of data replays. Our preliminary experiments reveal a consistent memorization-driven regime: performance continues to improve with repetition far beyond current practice (e.g., the commonly cited four-epoch limit). While a full scheduler remains future work, these insights provide a foundation for memorization-aware training schedules, helping to determine reuse budgets and move toward training LLMs smarter rather than longer with limited high-quality data.
arXiv:2607.04973v1 Announce Type: new Abstract: A thermal convection apparatus has been designed to study turbulent super structures at high Rayleigh numbers and Prandtl numbers of the order of unity. This apparatus consists of a rectangular cell with a length of $3.50\,\mathrm{m}$, width of $0.35\,\mathrm{m}$, and variable height, which is fixed at $0.70\,\mathrm{m}$ for the present study. This cell is installed inside a $5.6\,\mathrm{m}$ long pressure vessel facility, known as \emph{G\"ottingen Uboot}, which can be filled with compressed gasses (air, helium, nitrogen, or sulfur hexafluoride) at pressures up to $19\,\mathrm{bar}$, enabling Rayleigh numbers up to $ Ra \leq 5\times 10^{12}$ and Prandtl numbers of approximately $0.7 \leq Pr \leq 0.9$. The convection cell is bounded vertically by top and bottom plates consisting of a three-layer composite structure in which a thin Lexan plate is sandwiched between highly conductive aluminum plates. This allows for spatially resolved heat flux measurements. Each plate is subdivided into four longitudinal segments that can be independently temperature-controlled to enable homogeneous temperatures and the imposition of horizontal temperature gradients at both the top and bottom boundaries. While the bottom plate is electrically heated, the top plate's temperature is regulated using temperature-controlled circulating pressurized water. The apparatus is well suited for precise heat flux measurements, with the results obtained being in good agreement with those previously reported in the literature.
arXiv:2607.04977v1 Announce Type: new Abstract: Accurately estimating the unknown target label distribution is the critical first step for adapting to label shift. This task, widely known as quantification or class prevalence estimation, has recently seen significant advances through continuous KDE-based methods which model the density of multiclass classifier posteriors. Posterior vectors might be regarded as compositional data, since they lie on the probability simplex. However, existing KDE-based quantifiers typically rely on Euclidean Gaussian kernels, which ignore simplex geometry and incorrectly assign probability mass outside its boundaries. We introduce a geometry-aware KDE model for multiclass quantification based on log-ratio representations and Aitchison geometry, together with a shrinkage regularization that improves robustness near the simplex boundary. Combined with a maximum-likelihood interpretation of KDE-based quantification, we derive both point-estimation and Bayesian inference procedures for class prevalences. Experiments on 42 datasets across tabular, text, and image domains show that the proposed method is competitive with state-of-the-art quantifiers, often improving over standard KDE-based baselines, while also yielding strong results among Bayesian quantification methods.
arXiv:2607.05018v1 Announce Type: new Abstract: The development, tuning and operation of Monte Carlo event generators beyond the LHC era require vast amount of resources. In this study we investigate the sustainability of these software with a containerized set of tools (named 77rev/propripy), by benchmarking the HIJING++ heavy-ion Monte Carlo event generator. We analyze the performance of various CPU architectures and show that by choosing the level of multithreading properly, the cost of event generation can be optimized. The presented approach can reduce the energy footprint of high-energy physics event generators and therefore alleviate the ever-increasing, ubiquitous computational challenges.
arXiv:2602.06941v3 Announce Type: replace Abstract: Large language models can recover mid-generation from task-misaligned activation steering, producing explicit verbal restarts (e.g., ``wait, that's not right'') and continuing on-topic even while the steering perturbation remains active. We term this Endogenous Steering Resistance (ESR). Using sparse autoencoder (SAE) latents to steer model activations, we find that Llama-3.3-70B exhibits explicit ESR at 3.8%, with smaller models from the Llama-3 and Gemma-2 families showing the explicit form less frequently. Two controls dissociate ESR into a detection event and a sustained-resistance component that conditioning on recent on-topic tokens does not fully explain. We identify 26 SAE latents through contrastive on-topic/off-topic search; zero-ablating them reduces the multi-attempt rate by 25%, with random-latent and held-out-prompt controls supporting specificity. ESR can also be deliberately enhanced through both meta-prompting and fine-tuning on synthetic self-correction examples. ESR has dual implications for safety: it could harden models against adversarial activation-space manipulation, but may equally interfere with beneficial steering-based interventions, since the model has no way to distinguish the two. Code is available at https://github.com/agencyenterprise/endogenous-steering-resistance.
arXiv:2602.13517v2 Announce Type: replace Abstract: Large language models (LLMs) have demonstrated impressive reasoning capabilities by scaling test-time compute via long Chain-of-Thought (CoT). However, recent findings suggest that raw token counts are unreliable proxies for reasoning quality: increased generation length does not consistently correlate with accuracy and may instead signal "overthinking," leading to performance degradation. In this work, we quantify inference-time effort by identifying deep-thinking tokens -- tokens where internal predictions undergo significant revisions in deeper model layers prior to convergence. Across four challenging mathematical and scientific benchmarks (AIME 24/25, HMMT 25, and GPQA-diamond) and a diverse set of reasoning-focused models (GPT-OSS, DeepSeek-R1, and Qwen3), we show that deep-thinking ratio (the proportion of deep-thinking tokens in a generated sequence) exhibits a robust and consistently positive correlation with accuracy, substantially outperforming both length-based and confidence-based baselines. Leveraging this insight, we introduce Think@n, a test-time scaling strategy that prioritizes samples with high deep-thinking ratios. We demonstrate that Think@n matches or exceeds standard self-consistency performance while significantly reducing inference costs by enabling the early rejection of unpromising generations based on short prefixes.
arXiv:2607.05355v1 Announce Type: new Abstract: Attribution scores increasingly identify which neuron rows of a language model matter for applications such as pruning, interpretability, and editing for safety, yet whether they identify causally important rows is rarely tested directly. We address this with two paired audits built on one-shot neuron-row zeroing. We first audit selectors at the language-modeling level: attribution methods substantially outperform activation and magnitude-based baselines at identifying dispensable rows across five LLMs. We then adapt the same intervention into a behavior test by driving it with a contrastive harmful-versus-benign signal; the attributed rows are sufficient to install refusal on hate and crime while keeping benign over-refusal low and preserving language model fluency, and specific in that layer-matched random controls at the same depths fail. Highly rank-stable selectors can be among the least causally valid. Refusal moreover lives in a redundant subspace, where different attribution methods install it through largely disjoint row sets, so the recovered edit is one realization of a sufficient set rather than a unique mechanism. Together, these findings show that rank-stability proxies miss the kinds of selector failures a direct causal audit can surface.%
arXiv:2607.05362v1 Announce Type: new Abstract: We study the problem of reconfiguring a crossing-free embedding of a graph on a surface, with edges represented as curves, into another crossing-free embedding of the same graph on the same surface with the same fixed vertex positions. In this process, we reroute one edge at a time while maintaining crossing-free intermediate embeddings. This problem was introduced by Ito et al. [TALG 2025], who showed that even if the graph is a matching of two edges, reconfiguration is not always possible in the plane, but is always possible on the torus. For matchings of two or more edges, they gave a necessary and sufficient condition for reconfigurable embeddings in the plane, but not on the torus. Our main result is that for matchings, trees and forests, reconfiguration is always possible on the torus, and consequently, on any orientable surface of genus at least one. In addition, we provide sufficient conditions for reconfiguration on orientable surfaces of genus at least one and in the projective plane. For more general graphs, we show that reconfiguration is not always possible.
arXiv:2607.01400v2 Announce Type: replace Abstract: Deep multimodal brain-encoding models now predict fMRI responses to naturalistic video with high accuracy; whether their predicted neural signals also forecast behavioral engagement is unknown. We run TRIBE, the winning model of the 2025 Algonauts challenge (Llama-3.2 + V-JEPA 2 + Wav2Vec-BERT), on 48 YouTube videos and reduce its predicted cortical response to a per-second engagement curve, the global field power. Correlated against each video's "most replayed" heatmap, a proxy for re-watch, it shows no evidence of prediction: the pooled position-controlled partial correlation is +0.058 (95% CI [-0.04, 0.15]; t(47)=1.21, p=0.23), and not above simple loudness/motion baselines. The raw correlation is also near zero; the moderate values for music videos are an onset-replay artifact. The null holds across six cortical-network readouts, value/salience ROIs, and a permutation test; a supervised leave-one-video-out probe appears to reach r=0.47 but collapses to a temporal-shape artifact under a proper position control. Running the probe on TRIBE's input streams reveals at most a small, borderline visual-stream signal (matched vs. mismatched p=0.004-0.06) and none in audio, text, or the predicted cortex. The inter-subject-correlation readout, the closest prior positive result, is unavailable from the subject-averaged released model, so we fit our own per-subject encoders on the Algonauts fMRI (validated in-domain at r=0.15 and cross-domain, Friends-to-film, at r=0.10); the predicted ISC still does not track re-watch (r=-0.04, p=0.34). We bound rather than merely fail to reject the null: a Bayes factor gives moderate evidence for it (BF01=3.2), an equivalence test excludes effects above r=0.14, and the target's split-half reliability (0.82; ceiling r=0.9) rules out a noisy-label artifact. We release code, a video-ID manifest, and a heatmap-acquisition method robust to YouTube's SABR streaming.
arXiv:2607.05062v1 Announce Type: new Abstract: Often, the easiest way to learn something is to have to use it for a purpose. This purpose can be playful: In 'Tempus fugit', the player takes on the role of a magician who has to defeat enemies by casting spells. The applicability of spells and enemy attacks depends on the truth of formulas in linear temporal logic with past with respect to a trace that the player gradually builds. So, whoever wants to save the realm from monsters has to learn to read logic formulas. This paper describes the small browser game and explains our design choices. We expose how game mechanics connect to linear temporal logic with past over finite traces, and how this can help players approach a daunting topic like formal logic.
arXiv:2607.05199v1 Announce Type: new Abstract: Physics reasoning fails structurally in small language models: an error at any step propagates forward, corrupting every inference that follows. Limited domain knowledge, hallucination under multi-step derivation, and distributional sensitivity compound this failure. We propose a step-level reward framework that identifies the first reasoning error, generates targeted structured feedback, and trains the model to revise its solution via policy gradient with KL regularization, without exposing it to ground truth solutions as generation targets. Unlike annotation-dependent step-level methods, no preference data construction is required and the external verifier operates exclusively at training time. Across five physics benchmarks, our framework delivers accuracy gains of 17-20% over CoT prompting and 10-16% over the strongest baseline, reduces calculation errors from 56.9% to 23.5%, and reduces miscomprehension errors from 22.3% to 12.0% in the best observed cases. Conceptual errors reduce from 89.7% to 68.7%, yet persist as the hardest failure mode across all conditions.
arXiv:2607.05224v1 Announce Type: new Abstract: Code-switching (CS), alternating languages within the same utterance, poses significant challenges for automatic speech recognition (ASR) due to limited CS training data. This paper applies an iterative pseudo-labeling training approach to CS-ASR for the first time, demonstrating its effectiveness in leveraging unlabeled data to improve CS-ASR performance. The approach comprises three phases: pseudo-label generation, two-stage bilingual model training, and iterative improvements. It begins by generating pseudo-labels from a large unlabeled corpus, creating a semi-supervised dataset. This dataset supports a two-stage training framework where the model is pre-trained and then fine-tuned on supervised CS data. Iterative refinements further enhance the model's accuracy in handling complex CS scenarios. Our approach significantly advances CS-ASR systems, achieving notable Mix Error Rate (MER) reductions on SEAME's devman (6.35%) and devsge (8.29%) subsets.
arXiv:2607.05047v1 Announce Type: new Abstract: Capacity expansion is a key tool for planning future energy systems. However, weather-dependent generation and long-duration storage result in problem sizes that exceed the computational limits of conventional interior-point solvers, making it impossible to plan renewable systems that are cost-efficient and reliable across a wide range of weather conditions. To tackle such large problems, this paper introduces the double interior-point regularization (DIP-set) for Benders Decomposition, combining the advantages of traversing the interior of the solution space while remaining close to a reference solution. We benchmark the method on a power-sector problem and an energy-system problem, varying problem size and the level of foresight during operations. Results demonstrate that DIP-set outperforms competing regularizations in all test cases. The speed-up increases with size, reaching 30-50% for the largest problems, which are the most critical for planning renewable systems and are too large for state-of-the-art methods. The key benefit of DIP-set is its ability to mitigate the sharp decrease in convergence as BD approaches the optimal solution.
arXiv:2607.05133v1 Announce Type: new Abstract: World Action Models (WAMs) have shown strong potential for improving action generalization in autonomous driving by using future video prediction as dense supervision for scene dynamics and temporal causality. However, it remains unclear which architecture better transfers video-modeling benefits to trajectory generation. Existing cascaded or dual-DiT designs separate video imagination from action prediction, weakening the transfer of video-learned world dynamics to the trajectory branch: the action model may still overfit dataset-specific driving priors, while the video model only indirectly regularizes planning. We propose UNIVERSE, a unified video-action model built upon a single mask-modulated Diffusion Transformer. By co-training future video latents and ego-trajectory tokens within shared generative parameters, UNIVERSE allows dense video supervision to directly shape trajectory denoising, leading to stronger cross-domain action generalization. To ensure causal validity and efficient deployment, we introduce a Modality-Decoupling Visibility Mask, which shares historical context across modalities while blocking mutual attention between future video and trajectory tokens. This prevents future-target leakage and enables trajectory-only inference by removing future-video denoising at test time, achieving a $4.3\times$ speedup over joint video-action rollout while maintaining comparable planning accuracy. The same model also supports video-only and joint video-action rollouts. Experiments show that UNIVERSE achieves 91.0 PDMS on NAVSIM (vs. 89.6 for the Two-DiT variant), and demonstrates strong zero-shot transfer to nuScenes and Bench2Drive without fine-tuning, while ablations confirm the importance of single-DiT unification, video co-training, and mask-based modality decoupling.
arXiv:2607.05137v1 Announce Type: new Abstract: SoPlasmaFoam is an open-source, multi-region plasma-dielectric solver built on OpenFOAM, integrated with the PETSc linear-algebra suite (CPU and GPU back-ends), the blastAMR adaptive-mesh-refinement library (hexahedral and polyhedral meshes), and the ROUND family of high-resolution convective schemes. It solves drift-diffusion-reaction transport for charged species, coupled self-consistently to Poisson's equation explicitly or semi-implicitly, with plasma and dielectric regions joined by a monolithic multi-domain coupling for arbitrary curved interfaces. This work makes three contributions. First, a systematic assessment of convective schemes on a stiff scalar-advection problem and the positive-streamer benchmark shows that Scharfetter-Gummel is stable but excessively diffusive on coarse meshes, while ROUNDF outperforms all tested TVD limiters and is recommended for streamer transport. Second, an analysis of Poisson-transport coupling shows that fixed-point correction loops critically control accuracy, that a semi-implicit Poisson formulation does not remove this requirement, and that coupling must be tightened even when Courant and dielectric-relaxation numbers are well below unity. Third, a drift-robust wall boundary condition acting on discretized matrix coefficients is introduced, remaining accurate in the drift-dominated limit where conventional mixed-boundary mappings fail. The solver is validated against a low-pressure DC glow discharge and the positive-streamer benchmark, and its multi-region capability is demonstrated on a nanosecond surface dielectric barrier discharge. Performance analysis confirms memory-bound finite-volume scaling and shows that with AMR the solver is competitive with the fastest reported plasma codes. The framework provides a modular foundation for multiphysics simulations in plasma-assisted combustion, plasma processing, and plasma-based flow control.
arXiv:2607.02616v1 Announce Type: cross Abstract: The Multi-Level Intermediate Representation (MLIR) framework has become a cornerstone for building extensible, domain-specific compilers, with the quantum computing community already leveraging it to model quantum programs and implement basic optimizations. However, computationally intensive tasks in the quantum compilation pipeline, such as quantum circuit mapping, remain underexplored within the MLIR ecosystem. This paper proposes an MLIR-native blueprint for these non-local, quantum-specific optimization routines by reimplementing a well-established, state-of-the-art mapping A* search algorithm for qubit routing and SWAP insertion. Our evaluation demonstrates that this approach not only integrates seamlessly into an MLIR-based quantum compiler collection but also surpasses previous non-MLIR solutions in both solution quality and runtime. The implementation is open-source and publicly available at https://github.com/munich-quantum-toolkit/core.
arXiv:2607.03201v1 Announce Type: cross Abstract: Long-form recordings (LFRs) of child-centered audio are ecologically valid sources for studying early language development, but three problems limit their use. First, LFR corpora are collected across sites with heterogeneous formats and consent structures, making cross-corpus use non-trivial. Second, without standardized benchmarks, assessing whether tools generalize across languages and conditions is hard. Third, ML workflows rarely respect privacy constraints governing sensitive child speech. This paper presents a framework addressing all three: a standardized collection of 27 child-centered datasets built with open-source tools (S1); a replicable pipeline for four speech-processing benchmarks (S2); and ELSI, a role-based ecosystem embedding ethical governance into the ML workflow (S3). We demonstrate the framework via a voice type classification case study and show the three solutions are mutually dependent.
arXiv:2504.06842v4 Announce Type: replace Abstract: We introduce the Gradient-MUSIC algorithm for estimating the unknown frequencies and amplitudes of a nonharmonic signal from noisy time samples. While the classical MUSIC algorithm performs a computationally expensive search over a fine grid, Gradient-MUSIC is significantly more efficient and eliminates the need for discretization over a fine grid by using optimization techniques. It coarsely scans the 1D landscape to find initialization simultaneously for all frequencies followed by parallelizable local refinement via gradient descent. We also analyze its performance when the noise level is sufficiently small and the signal frequencies are separated by at least $8\pi/m$, where $\pi/m$ is the standard resolution of this problem. Even though the 1D landscape is nonconvex, we prove a global convergence result for Gradient-MUSIC: coarse scanning provably finds suitable initialization and gradient descent converges at a linear rate. In addition to convergence results, we also upper bound the error between the true signal frequencies and amplitudes with those found by Gradient-MUSIC. For example, if the noise has $\ell^\infty$ norm at most $\varepsilon$, then the frequencies and amplitudes are recovered up to error at most $C\varepsilon/m$ and $C\varepsilon$ respectively for a universal $C>0$, which are minimax optimal in $m$, $\varepsilon$, and number of frequencies. Our theory can also handle stochastic noise with performance guarantees under nonstationary independent Gaussian noise. Our main approach is a comprehensive geometric analysis of the landscape, a perspective that has not been explored before.