arXiv subjects

Kevin Chen

Publications and source records attributed to Kevin Chen.

At least 19 recordsLinked to original sources

arXiv2026

WASP-Mem: Let Agent Organize Procedural Memory. Human Spend Effort on Diagnosis

LLM agents struggle to automate complex enterprise procedures because static, pre-built harness cannot capture proprietary policies, undocumented behaviors, or state-dependent recovery paths that surface only during execution. Procedural memory accrued from agent experience offers a training-free path to closing this gap -- yet unlike episodic records or semantic facts, it must consolidate repeated successes and diagnosed failures into reusable actions and recovery strategies. We introduce WASP-Mem (Wiki-based Agent-Synthesized Procedural Memory), a nimble but effective method that distills execution traces and feedback into persistent, interlinked wiki pages, enabling continual agent learning. On TheAgentCompany benchmark, WASP-Mem raises the checkpoint pass rate by 4.6-6.1% over the strongest baseline we compared across three models while requiring 57-61% less storage. We further present a systematic controlled study spanning consolidation method, feedback type, trace ingestion scope, and refinement rounds. Our results reveal that prescriptive knowledge organization such as human defined schemas and explicit deduplication rules hurts performance rather than closing the procedural-knowledge gap. In contrast, human diagnosis converts failed traces into corrective knowledge and drives sustained improvement: a single round of human diagnosis outperforms four rounds of self-judgment by 7.0-10.9%. These findings point to a simple division of labor: let agents organize procedural memory, and invest human effort where it matters most -- diagnosis.

cs.AI↗

arXiv2026

Safe by Benchmark, Unsafe in Practice: Lessons from 1331 Child-Related AI Incidents

AI systems are increasingly present in children's everyday environments: education, entertainment, social networks, and toys. Their safety is typically assessed at the model level, through benchmarks, red-teaming, and content filtering. However, most AI harm taxonomies lack child-specific evidence and under-specify developmental harm and the stakeholder asymmetries unique to minors. Building on prior taxonomies, we construct a child-centered harm taxonomy mapped to sociotechnical failure modes. We curate and code 1331 publicly reported child-related AI incidents, annotating internal and external harm types and the failure mode implicated in each incident. Our analysis reveals which harms are most prevalent in public reporting, how individual-level and population-level harms relate, and which failures are most often coded alongside harm. We find that harm concentrates in representational, dignity-related, and psychological categories, and co-occurs most often with deployment-context, governance, and misuse failures rather than model defects. Harm to an individual child is also frequently linked to population-level harm, reaching children who never interacted with the system. These findings suggest that model-level evaluations, while necessary, cannot by themselves anticipate the harm pathways most often documented for children. We conclude with actionable implications emphasizing mitigations beyond model-level safeguards such as content filtering: age assurance, consent verification for depiction, safety-by-design UX constraints, incident disclosure norms, and auditing regimes targeting developmental risk. Privacy-minimized annotations, a codebook, and documentation are available at https://github.com/The-Responsible-AI-Initiative/KidAID.

cs.AI↗

arXiv2026

A Process-Informed, Photorealistic Digital Twin for Vision-Based Inline Inspection of C-SMC Strip Deposition

Inline quality assurance for Sheet Molding Compound (SMC) manufacturing is hindered by the lack of large-scale, accurately labeled image datasets for training machine-learning models. This work presents a process-informed, photorealistic digital twin for carbon-fiber SMC strip deposition. A rigid-body simulation reproduces the deposition kinematics (settling, friction, belt motion), while a physically grounded polarization-rendering model produces realistic appearance, so that every image carries exact, mutually consistent labels in synchronized formats: oriented bounding boxes, point annotations, density maps, and polarization renderings (Angle and Degree of Linear Polarization). Because the joint distribution of many overlapping strips cannot be specified analytically, simulation yields exact, scalable ground truth that manual annotation cannot provide. The twin produces over 16,000 images across 3,300 frames, and the synthetic data replicate near-isotropic fiber orientation (axial concentration r < 0.05) consistent with as-deposited commercial SMC. Trained purely on synthetic data, YOLOv11 and CSRNet reach detection mAP50 up to 0.91, with a dedicated oriented-bounding-box detector confirming the oriented labels are directly trainable (mAP50 = 0.83); on real manufacturing images, the synthetic-to-real difference in edge and texture structure (Kolmogorov-Smirnov distance 0.46, distribution overlap 0.56) is of the same magnitude as the difference between two real acquisition series of the same material (0.41, overlap 0.57). The approach substantially reduces the labeling burden for camera-based inline inspection of fiber orientation and volume content.

cs.CV↗

arXiv2026

Structured Latent Dynamics for Long-Horizon Embodied World Models

Reliable long-horizon prediction remains a central challenge for embodied world models. Although recent generative models can synthesize visually plausible action-conditioned futures, their latent dynamics are typically learned through unconstrained transition modules, which may gradually deviate from coherent physical evolution as predictions are recursively rolled forward. We investigate whether explicitly structuring latent state evolution can improve the stability and utility of such imagined trajectories. To this end, we introduce HWM, a structured dynamics framework that represents backbone latents in a learned phase space consisting of generalized coordinates and momenta. Future states are propagated through dissipative Hamiltonian dynamics and mapped back to the original latent representation through a learnable inverse interface. The resulting formulation can be integrated into existing world-model architectures without redesigning their perception or decoding components, while multi-step transition supervision encourages the structured rollout to remain aligned with the latent trajectories induced by real observations. Controlled simulation experiments show that HWM produces more accurate estimates of velocity, acceleration, and jerk and substantially reduces long-horizon energy drift relative to unstructured latent dynamics. We further evaluate the approach across three distinct embodied world-model settings, covering single-arm multi-view prediction with Ctrl-World, dual-arm single-view prediction with EVAC, and task-oriented single-arm prediction with Cosmos-Predict2.5-2B. On RoboCasa, incorporating HWM into Cosmos-Predict2.5-2B yields a 2.7% relative improvement in task success. On real-world AgiBot World trajectories, it improves nDTW by 9.7% with Ctrl-World and Logic by 8.4% with EVAC.

cs.LG↗

arXiv2026

Output-aware Residual Stream Pruning for Large Language Models

Residual stream pruning methods reduce inference cost by shrinking the model's hidden dimension, but existing approaches typically choose these dimensions by minimizing activation reconstruction error. This criterion implicitly treats all perturbation directions as equally important, ignoring the sensitivity of downstream layers. We introduce a sensitivity-aware approach to residual-stream pruning that directly accounts for this direction-dependent sensitivity. Using a second-order approximation to the output KL divergence, we characterize the effect of a residual-stream perturbation through both its activation covariance and the local sensitivity of the model output. The resulting subspace selection objective couples these two quantities, but is difficult to optimize directly. We derive a tractable spectral upper bound that reduces subspace selection to an eigendecomposition of a sensitivity-weighted covariance matrix, retaining the efficiency and structural simplicity of rotation-based pruning methods. Across several instruction-tuned language model families, our method consistently reduces calibration KL divergence relative to activation-only pruning and improves perplexity and downstream task performance over a range of compression levels. Our results show that preserving activation energy alone is insufficient for residual-stream pruning, and that explicitly accounting for how perturbations propagate to the model output provides a more effective criterion for selecting dimensions to remove.

cs.LG↗

arXiv2026

Three-Wave Mixing as a High-Throughput Diagnostic Tool for Nonlinear Waveguides

The recent development of wavelength-scale nonlinear waveguides highlights the need for new diagnostic methods that characterize both waveguide dispersion and fabrication-induced spatial inhomogeneities. While tightly confining waveguides enable compact and highly efficient nonlinear devices, nanometer-scale waveguide nonuniformities can limit the effective interaction length and conversion efficiency. In this work, we establish a high-throughput diagnostic method for second-order nonlinear waveguides based on near-degenerate three-wave mixing, hereafter referred to as sum-frequency tomography (SFT). In contrast with conventional diagnostics based on second-harmonic generation, SFT produces two-dimensional transfer functions that allow dispersion orders to be characterized separately from longitudinal phase-mismatch variations. We validate this technique using commercially available PPLN waveguides and show that SFT recovers the group-velocity mismatch between the interacting waves, the group-velocity dispersion of the fundamental, and a low-order polynomial approximation of the longitudinal phase-mismatch profile. SFT avoids the experimental complexities of phase-sensitive or spatially resolved measurements, enabling the characterization of individual waveguides in seconds.

physics.optics↗

arXiv2026

2D target search with boundary-induced resetting

Most studies of stochastic search processes with resetting focus on spontaneous resetting events that occur at a random sequence of times typically generated by a Poisson process. One of the major consequences of spontaneous resetting is that it yields a finite mean first passage time (MFPT) in unbounded domains, resulting in an optimal resetting rate that minimises the MFPT. An optimal resetting rate can also occur in bounded domains provided that the reset point is not too far from the target. Recently, there has been growing interest in event-based rather than spontaneous resetting, where resetting is triggered when a specific threshold is reached. Identifying the threshold with a physical non-target boundary then leads to so-called boundary-induced resetting. In this paper, we explore the effects of boundary-induced resetting on single-particle diffusive search in a two-dimensional (2D) bounded domain $Ω$ containing one or more small interior targets. The interior target boundaries are totally absorbing, whereas the exterior boundary $\partial Ω$ is sticky. That is, whenever the particle reaches a point on $\partial Ω$ it remains attached for a random waiting time $τ$ after which it immediately resets to a fixed interior point $\x_0\in Ω$. Using renewal theory, matched asymptotic analysis and Green's functions we calculate the unconditional MFPT for absorption by any of the targets and compare this with the corresponding MFPT of a search process on the same domain with a reflecting surface $\partial Ω$ and no resetting. We show that boundary-induced resetting only reduces the MFPT relative to the no resetting case if the searcher's reset point $\x_0$ is sufficiently close to one of the targets.

cond-mat.stat-mech↗

arXiv2026

Full-Stack High-Volume Quantum Networking Architecture based on Photonic-Integrated Tin Vacancy Centers in Diamond

Solid state quantum emitters are a leading platform for photonic quantum networking with memory nodes. However, the inhomogeneous distribution of quantum emitters, as well as several environmental factors (i.e. strain and electric fields) spread the frequency spectrum of the qubits, making them distinguishable and therefore not a reliable resource for distributed quantum entanglement. In this paper, we demonstrate a full-stack approach to integrating nearly indistinguishable tin vacancy (SnV$^-$) quantum emitters on a frequency-tunable photonic interposer that overcomes the native distribution and static variation of quantum emitters for an indistinguishable photonic quantum networking platform. We demonstrate a silicon nitride-on-insulator photonic integrated circuit (PIC) with accompanying multiphysics digital twin (MPhDT) that guides discovery of SnV$^-$ strain-tuning parameters and informs construction of a multi-channel quantum repeater node. On this node, we achieve the first simultaneous demonstration of spectral tuning of the zero phonon line (ZPL) at GHz scale; coherent electron spin control with gate times of $<80$ ns; strongly- and weakly-coupled nuclear spin detection; and commercial fiber array-coupled readout of a SnV$^-$ center. Finally, we propose and simulate improvements to the architecture that achieve 99.96% connectivity of $ N \sim 1000$ emitters spanning the inhomogeneous distribution of SnV$^-$ centers in strained diamond, where distributed quantum entanglement may be realized.

quant-ph↗

arXiv2026

LAWFUL: Law-Aligned Witness for Faithful Use of Latents

When a neural network predicts a physical system accurately, has it learned the governing law as formal, structured knowledge, and if so, does the network's internal computation actually use that representation throughout the law's domain of validity? We identify four interpretability gaps that limit answering these questions for {\em physics laws over continuous variables}: the absence of a coverage-aware causal-consistency measure over continuous counterfactuals; of a domain-of-validity test for the identified circuit; of a verification of the law's invariants and forbidden behaviors; and of a quantification of how a derived physical quantity flows through the circuit. We develop a foundational framework, LAWFUL, that closes the first two and lays groundwork for the remaining two, and illustrate it on the Mocap2Radar transformer, validating whether it learns and internally uses the Doppler frequency law $f(t) = \frac{2 v(t)}λ$ from motion-capture and radar data in which neither $f(t)$ nor $v(t)$ appears.

cs.LG↗

arXiv2026

From Text to Discovery: How Large Language Models Are Reshaping Research Across Scientific and Humanistic Disciplines

Large Language Models (LLMs) are rapidly reshaping academic research across the natural sciences, social sciences, and humanities, yet the scientific community lacks a comprehensive, cross-disciplinary account of how these tools are being integrated, what they deliver, and where they fall short. This paper addresses that gap by mapping their current state and outlining an agenda for their responsible integration into scientific research. Our analysis reveals a consistent pattern: LLMs meaningfully accelerate research workflows -- from hypothesis generation and literature synthesis to data analysis and scientific writing -- while introducing serious challenges related to hallucination, reproducibility, dataset bias, and model opacity. Beyond technical limitations, we identify ten underexplored challenges, including the erosion of researcher autonomy, AI-driven confirmation bias, authorship ambiguity, and unequal access to these technologies -- systemic risks that demand interdisciplinary governance frameworks, robust validation standards, and expanded explainability research.

cs.DL↗

arXiv2026

LLM Harms: A Taxonomy and Discussion

This study addresses categories of harm surrounding Large Language Models (LLMs) in the field of artificial intelligence. It addresses five categories of harms addressed before, during, and after development of AI applications: pre-development, direct output, Misuse and Malicious Application, and downstream application. By underscoring the need to define risks of the current landscape to ensure accountability, transparency and navigating bias when adapting LLMs for practical applications. It proposes mitigation strategies and future directions for specific domains and a dynamic auditing system guiding responsible development and integration of LLMs in a standardized proposal.

cs.CY↗

arXiv2026

LLMs and Childhood Safety: Identifying Risks and Proposing a Protection Framework for Safe Child-LLM Interaction

Large Language Models (LLMs) are increasingly embedded in child-facing contexts such as education, companionship, creative tools, but their deployment raises safety, privacy, developmental, and security risks. We conduct a systematic literature review of child-LLM interaction risks and organize findings into a structured map that separates (i) parent-reported concerns, (ii) empirically documented harms, and (iii) gaps between perceived and observed risk. Moving beyond descriptive listing, we compare how different evidence streams in surveys, incident reports, youth interaction logs, and governance guidance operationalize "harm," where they conflict, and what mitigations they imply. Based on this synthesis, we propose a protection framework that couples child-specific content safety and developmental sensitivity with security-grade controls for adversarial misuse, including prompt injection and multimodal jailbreak pathways. The framework specifies measurable evaluation targets (e.g., harmful-content avoidance, age-calibrated readability, bias parity checks, prompt-injection robustness, and monitoring transparency) to support developers, educators, and policymakers in assessing and improving child-safe LLM deployments.

cs.CY↗

arXiv2026

Safe-Child-LLM: A Developmental Benchmark for Evaluating LLM Safety in Child-LLM Interactions

As Large Language Models (LLMs) increasingly power applications used by children and adolescents, ensuring safe and age-appropriate interactions has become an urgent ethical imperative. Despite progress in AI safety, current evaluations predominantly focus on adults, neglecting the unique vulnerabilities of minors engaging with generative AI. We introduce Safe-Child-LLM, a comprehensive benchmark and dataset for systematically assessing LLM safety across two developmental stages: children (7-12) and adolescents (13-17). Our framework includes a novel multi-part dataset of 200 adversarial prompts, curated from red-teaming corpora (e.g., SG-Bench, HarmBench), with human-annotated labels for jailbreak success and a standardized 0-5 ethical refusal scale. Evaluating leading LLMs -- including ChatGPT, Claude, Gemini, LLaMA, DeepSeek, Grok, Vicuna, and Mistral -- we uncover critical safety deficiencies in child-facing scenarios. This work highlights the need for community-driven benchmarks to protect young users in LLM interactions. To promote transparency and collaborative advancement in ethical AI development, we are publicly releasing both our benchmark datasets and evaluation codebase at https://github.com/The-Responsible-AI-Initiative/Safe_Child_LLM_Benchmark.git

cs.CY↗

arXiv2026

What Physics do Data-Driven MoCap-to-Radar Models Learn?

Data-driven MoCap-to-radar models generate plausible micro-Doppler spectrograms, but do they actually learn the underlying physics? We introduce a physics-based interpretability framework to answer this question via two proposed complementary metrics: one measures alignment between model predictions and the physics-derived Doppler frequency, while the other tests whether predictions preserve the velocity-frequency relationship under velocity intervention. Both metrics require only MoCap input and model predictions, without access to measured radar data. Experiments across several model architectures reveal that low reconstruction error does not guarantee physical consistency: some, but not all, models achieve low error yet perform poorly on the two physics-based metrics. Further analysis shows that temporal attention is critical for transformer-based models to learn the underlying physics.

cs.LG↗

arXiv2026

Entropy-Preserving Reinforcement Learning

Policy gradient algorithms have driven many recent advancements in language model reasoning. An appealing property is their ability to learn from exploration on their own trajectories, a process crucial for fostering diverse and creative solutions. As we show in this paper, many policy gradient algorithms naturally reduce the entropy -- and thus the diversity of explored trajectories -- as part of training, yielding a policy increasingly limited in its ability to explore. In this paper, we argue that entropy should be actively monitored and controlled throughout training. We formally analyze the contributions of leading policy gradient objectives on entropy dynamics, identify empirical factors (such as numerical precision) that significantly impact entropy behavior, and propose explicit mechanisms for entropy control. These include REPO, a family of algorithms that modify the advantage function to regulate entropy, and ADAPO, an adaptive asymmetric clipping approach. Models trained with our entropy-preserving methods maintain diversity throughout training, yielding final policies that are more performant and retain their trainability for sequential learning in new environments.

cs.LG↗

arXiv2026

RAGPerf: An End-to-End Benchmarking Framework for Retrieval-Augmented Generation Systems

We present the design and implementation of a RAG-based AI system benchmarking (RAGPerf) framework for characterizing the system behaviors of RAG pipelines. To facilitate detailed profiling and fine-grained performance analysis, RAGPerf decouples the RAG workflow into several modular components - embedding, indexing, retrieval, reranking, and generation. RAGPerf offers the flexibility for users to configure the core parameters of each component and examine their impact on the end-to-end query performance and quality. RAGPerf has a workload generator to model real-world scenarios by supporting diverse datasets (e.g., text, pdf, code, and audio), different retrieval and update ratios, and query distributions. RAGPerf also supports different embedding models, major vector databases such as LanceDB, Milvus, Qdrant, Chroma, and Elasticsearch, as well as different LLMs for content generation. It automates the collection of performance metrics (i.e., end-to-end query throughput, host/GPU memory footprint, and CPU/GPU utilization) and accuracy metrics (i.e., context recall, query accuracy, and factual consistency). We demonstrate the capabilities of RAGPerf through a comprehensive set of experiments and open source its codebase at GitHub. Our evaluation shows that RAGPerf incurs negligible performance overhead.

cs.PF↗

arXiv2026

Generative AI and LLMs in Industry: A text-mining Analysis and Critical Evaluation of Guidelines and Policy Statements Across Fourteen Industrial Sectors

The rise of Generative AI (GAI) and Large Language Models (LLMs) has transformed industrial landscapes, offering unprecedented opportunities for efficiency and innovation while raising critical ethical, regulatory, and operational challenges. This study conducts a text-based analysis of 160 guidelines and policy statements across fourteen industrial sectors, utilizing systematic methods and text-mining techniques to evaluate the governance of these technologies. By examining global directives, industry practices, and sector-specific policies, the paper highlights the complexities of balancing innovation with ethical accountability and equitable access. The findings provide actionable insights and recommendations for fostering responsible, transparent, and safe integration of GAI and LLMs in diverse industry contexts.

cs.CY↗

arXiv2026

Evaluating LLM Safety Across Child Development Stages: A Simulated Agent Approach

Current safety alignment for Large Language Models (LLMs) implicitly optimizes for a "modal adult user," leaving models vulnerable to distributional shifts in user cognition. We present ChildSafe, a benchmark that quantifies alignment robustness under cognitive shifts corresponding to four developmental stages. Unlike static persona-based evaluations, we introduce a parametric cognitive simulation approach, formalizing developmental stages as hyperparameter constraints (e.g., volatility, context horizon) to generate out-of-distribution interaction traces. We validate these agents against ground-truth human linguistic data (CHILDES) and deploy them across 1,200 multi-turn interactions. Our results reveal a systematic alignment generalization gap: state-of-the-art models exhibit up to 11.5% performance degradation when interacting with early-childhood agents compared to standard baselines. We provide the research community with the validated agent artifacts and evaluation protocols to facilitate robust alignment testing against non-adversarial, cognitively diverse populations.

cs.CY↗