arXiv subjects

Michael Nguyen

Publications and source records attributed to Michael Nguyen.

10 recordsLinked to original sources

arXiv2026

Where Does Harness-Optimization Value Live? Localized Gains and the Budget-Splitting Trap in Self-Evolving LLM Agents

A growing body of work improves frozen large language models (LLMs) as agents by evolving their harness: the textual scaffolding around the model, including persona, strategy, format rules, and control heuristics. Existing reflective prompt-evolution methods usually optimize this harness as one flat string. We instead ask where the optimization value actually resides. We introduce HARNESSEVO, which decomposes the harness into four separately evolvable slots: role, task-strategy, tool/format-rules, and reflection/control. Using the same reflective optimizer under an iso-budget setting, we pair this decomposition with leave-one-in and leave-one-out attribution to measure the contribution of each slot. On ALFWorld with a frozen 7B backbone, HARNESSEVO does not significantly improve the overall binary success rate over either the stock harness or flat-string evolution: 0.657 versus 0.642 and 0.642, respectively. However, the slot-level analysis reveals that nearly all useful optimization value is localized in the reflection/control slot, which achieves a leave-one-in gain of +0.119. The other slots are individually null. We further show that uniform budget splitting is harmful: allocating 64 rollouts across four slots leaves only 16 per slot, below the optimizer's effective search floor, causing every slot to freeze at its empty seed. Concentrating the budget on the high-credit control slot recovers the lost gain, reaching 0.761 with half the split budget. The effect is task-contingent. On WebShop, all slots freeze empty and all methods tie, indicating a genuine absence of recurrent, verbalizable control failures rather than budget starvation. Overall, our results suggest that harness value is localized, uniform budget splitting can be actively harmful, and credit assignment should precede structured agent-evolution.

cs.CL↗

arXiv2026

Recursive Self-Evolving Agents via Held-Out Selection

LLM agents are increasingly improved without weight updates by evolving a natural-language artifact, such as reflections, workflows, playbooks, cheatsheets, or optimized prompts, that conditions a frozen policy. Such methods are typically reported as wins on the single benchmark where they help. We study them apples-to-apples and surface a sharper picture. We introduce RSEA, a Recursive Self-Evolving Agent that carries a compact three-layer natural-language state: an imperative strategy, reusable skills, and a procedural playbook. Across generations, RSEA rewrites all three layers from its own trajectories and commits a candidate only if it does not regress on a disjoint held-out split, using a strict keep-better gate. Across four diverse benchmarks, ALFWorld, GAIA, (\tau)-bench, and WebShop, and six faithful baselines, ReAct, Reflexion, GEPA, AWM, ACE, and Dynamic Cheatsheet, all evaluated on one shared local backbone, we find three main results. First, no artifact universally wins. RSEA is the strongest single-pass method on ALFWorld, reaching 69.3% compared with 64.6% for ReAct (McNemar (p=0.015)), and reaches 79.4% with retry, the best overall result. However, concrete-workflow induction, represented by AWM, is best on the strong-backbone tool-use tasks. Second, unguarded context evolution is high-variance and unsafe. Dynamic Cheatsheet, which curates context online without a held-out gate, is near-best on ALFWorld at 70.7%, yet collapses on WebShop, with a score of 0.14 compared with 0.43 for ReAct. Third, RSEA's strict held-out selection is what makes recursive self-evolution monotone-safe: it never significantly underperforms the base agent on any benchmark and falls back to vanilla ReAct when evolved context would hurt.

cs.AI↗

arXiv2025

FRIT: Using Causal Importance to Improve Chain-of-Thought Faithfulness

Chain-of-thought (CoT) reasoning has emerged as a powerful tool for improving large language model performance on complex tasks, but recent work shows that reasoning steps often fail to causally influence the final answer, creating brittle and untrustworthy outputs. Prior approaches focus primarily on measuring faithfulness, while methods for systematically improving it remain limited. We introduce Faithful Reasoning via Intervention Training (FRIT), a scalable alignment method that trains models to produce causally consistent reasoning by learning from systematically corrupted examples. FRIT generates synthetic training data by intervening on individual reasoning steps in model-generated CoTs, creating faithful/unfaithful pairs that highlight when reasoning breaks down. We then apply Direct Preference Optimization to teach models to prefer causally consistent reasoning paths. Evaluating on Qwen3-8B and Mistral-7B-v0.1 across factual and symbolic reasoning tasks, FRIT increases faithful reasoning by $3.4$ percentage points for Mistral on GSM8K while improving accuracy by $7.6$ percentage points. Our approach provides the first scalable, supervision-free method for training language models to produce more reliable and interpretable reasoning, addressing a critical gap between reasoning performance and trustworthiness. We release our code at \href{https://github.com/Anut-py/frit}.

cs.AI↗

arXiv2025

Renormalized mechanics and stochastic thermodynamics of growing vesicles

Uncovering the rules governing the nonequilibrium dynamics of the membranes that define biological cells is of central importance to understanding the physics of living systems. We theoretically and computationally investigate the behavior of flexible quasispherical vesicles that exchange membrane constituents, internal volume, and heat with an external reservoir. The excess chemical potential and osmotic pressure difference imposed by the reservoir act as generalized thermodynamic driving forces that modulate vesicle morphology. We show that the renormalization of membrane mechanical properties by nonequilibrium driving gives rise to a morphological transition between a weakly driven regime, in which growing vesicles remain quasispherical, and a strongly driven regime, in which vesicles accommodate rapid membrane uptake by developing surface wrinkles. Additionally, we propose a minimal vesicle growth-shape law, derived using insights from stochastic thermodynamics, that robustly describes vesicle growth dynamics even in strongly driven, far-from-equilibrium regimes.

cond-mat.soft↗

arXiv2025

Deep Learning-Powered Classification of Thoracic Diseases in Chest X-Rays

Chest X-rays play a pivotal role in diagnosing respiratory diseases such as pneumonia, tuberculosis, and COVID-19, which are prevalent and present unique diagnostic challenges due to overlapping visual features and variability in image quality. Severe class imbalance and the complexity of medical images hinder automated analysis. This study leverages deep learning techniques, including transfer learning on pre-trained models (AlexNet, ResNet, and InceptionNet), to enhance disease detection and classification. By fine-tuning these models and incorporating focal loss to address class imbalance, significant performance improvements were achieved. Grad-CAM visualizations further enhance model interpretability, providing insights into clinically relevant regions influencing predictions. The InceptionV3 model, for instance, achieved a 28% improvement in AUC and a 15% increase in F1-Score. These findings highlight the potential of deep learning to improve diagnostic workflows and support clinical decision-making.

eess.IV↗

arXiv2024

Conformer-1: Robust ASR via Large-Scale Semisupervised Bootstrapping

This paper presents Conformer-1, an end-to-end Automatic Speech Recognition (ASR) model trained on an extensive dataset of 570k hours of speech audio data, 91% of which was acquired from publicly available sources. To achieve this, we perform Noisy Student Training after generating pseudo-labels for the unlabeled public data using a strong Conformer RNN-T baseline model. The addition of these pseudo-labeled data results in remarkable improvements in relative Word Error Rate (WER) by 11.5% and 24.3% for our asynchronous and realtime models, respectively. Additionally, the model is more robust to background noise owing to the addition of these data. The results obtained in this study demonstrate that the incorporation of pseudo-labeled publicly available data is a highly effective strategy for improving ASR accuracy and noise robustness.

eess.AS↗

arXiv2021

A mean field thermodynamic framework for time dependent self-assembly and pattern formation

In this work, we use a minimal model to introduce a framework for controlling self-assembly under the influence of time-dependent driving forces. We develop a mean-field thermodynamic framework that predicts the conditions required to reliably self-assemble a desired spatial pattern under time-varying external fields. We also calculate the entropy production associated with the time-dependent self-assembly process and examine how it can be used to predict conditions under which the external time-varying signal is reliably encoded as a spatial pattern in the self-assembling material. While the results in this work are developed in the context of a minimal one-dimensional model, we anticipate that the framework can be used to establish guidelines for controlling self-assembly in more complex scenarios.

cond-mat.soft↗

arXiv2020

A strong non-equilibrium bound for sorting of crosslinkers on growing biopolymers

Understanding the role of non-equilibrium driving in self-organization is crucial for developing a predictive description of biological systems, yet it is impeded by their complexity. The actin cytoskeleton serves as a paradigm for how equilibrium and non-equilibrium forces combine to give rise to self-organization. Motivated by recent experiments that show that actin filament growth rates can tune the morphology of a growing actin bundle crosslinked by two competing types of actin binding proteins, we construct a minimal model for such a system and show that the dynamics are subject to a set of thermodynamic constraints that relate the non-equilibrium driving, bundle morphology, and molecular fluxes. The thermodynamic constraints reveal the importance of correlations between these molecular fluxes, and offer a route to estimating microscopic driving forces from microscopy experiments.

cond-mat.soft↗

arXiv2018

Dissipation induced transitions in two dimensional elastic membranes

Stochastic thermodynamics provides a useful set of tools to analyze and constrain the behavior of far from equilibrium systems. In this paper, we report an application of ideas from stochastic thermodynamics to the problem of membrane growth. Non-equilibrium forcing of the membrane can cause it to buckle and undergo a morphological transformation. We show how ideas from stochastic thermodynamics, in particular the recently derived thermodynamic uncertainty relations, can be used to phenomenologically describe and constrain the parameters required to excite morphological changes during a non-equilibrium growth process.

cond-mat.soft↗

arXiv2015

Design principles for non-equilibrium self-assembly

We consider an important class of self-assembly problems and using the formalism of stochastic thermodynamics, we derive a set of design principles for growing controlled assemblies far from equilibrium. The design principles constrain the set of structures that can be obtained under non-equilibrium conditions. Our central result provides intuition for how equilibrium self-assembly landscapes are modified under finite non-equilibrium drive.

cond-mat.stat-mech↗