arXiv subjects

Wang

Publications and source records attributed to Wang.

At least 19 recordsLinked to original sources

arXiv2026

Mitra-v2 Technical Report

We introduce Mitra-v2, a tabular foundation model that delivers state-of-the-art performance on real-world classification and regression problems, from credit-risk scoring and clinical prediction to equipment-failure detection and house-price estimation. Mitra-v2 is trained only on synthetic data, with a pretraining distribution that is much larger and more diverse than Mitra-v1's. Built on a small 2D Transformer backbone, Mitra-v2 supports longer contexts and larger feature spaces. Improved optimization lets it learn from this larger task distribution. We evaluate Mitra-v2 on the TabArena and TALENT benchmarks, comprising more than 300 real-world datasets under two evaluation protocols. On the full TabArena benchmark, Mitra-v2 delivers state-of-the-art performance at the level of the industry-scale TabFM and EXAONE Tabular models, while surpassing TabPFN-3 by a wide margin in both classification and regression. Mitra-v2 matches the 1.6B-parameter TabFM with only 5% of its size (77M parameters), delivering frontier performance at a fraction of the cost. On TALENT, Mitra-v2 remains among the leading models, clearly outperforming TabPFN-3 and TabICLv2. It also ranks first on classification tasks with more than ten classes, even though it was pretrained only on tasks with at most ten classes. These results make Mitra-v2 one of the strongest and most broadly applicable open tabular foundation models released to date. We release the model weights, the inference and fine-tuning code, and our evaluation results under the Apache-2.0 license.

cs.LG↗

arXiv2026

@skills: Attention is all you have

There are 56,804 public agent skills today, and teams write many more privately. The dominant delivery model is installation: once installed, a skill's description remains in the system prompt, competing for fewer than 100 reliable trigger slots. This leaves the long tail with no practical path to use and forces teams' own playbooks to compete for the same scarce space. We observe that installation bundles three separable functions: content, persistence, and automatic triggering. Only the last requires prompt residency. We therefore propose @skills, an open protocol that separates them. A path addresses any skill, subtree, or collection, and reading a skill is sufficient to use it, so nothing is installed or made resident. The operation vendors a copy at the same path into a project's Git-tracked tree for adaptation and ownership. The operation adds one .gitignore-style line, the only element that costs prompt residency. A directory is a menu, making bundles ordinary directories rather than all-or-nothing units. The protocol requires no manifest, lockfile, or registration, and SKILL.md remains unchanged. @skills is additive, ships as an installable package, and turns any agent that can read files and run commands into a client through a single instruction file. Its open specification is at https://github.com/SylphAI-Inc/atskills and it is implemented in the AdaL CLI at https://adalagent.ai . Because paths address skills well but cannot find them, the protocol is paired with a free hub at https://atskills.one for corpus-wide search and ranking, repository-free hosting, private and team collections, and one-screen authoring. The hub is optional: gh: and local paths resolve without it, and indexed GitHub skills retain their gh: identities. Install less, use more.

cs.AI↗

arXiv2026

Muse Spark Safety & Preparedness Report

Muse Spark is the latest large language model developed by Meta. In this report, we first present evaluations for catastrophic risk domains under Meta's Advanced AI Scaling Framework, along with the evidence that informed our launch decision. We then discuss additional considerations, such as Muse Spark's broader content safety and behavioral profile, that are relevant to overall safety but fall outside the catastrophic risk domains governed by the Framework. Our preparedness results covering Chemical and Biological, Cybersecurity, and Loss of Control risks assess Muse Spark's deployment within Meta AI as presenting acceptable levels of residual risks under our Advanced AI Scaling Framework. We conducted a broad set of evaluations targeting dual-use and high-risk capabilities across these catastrophic risk domains. Those evaluations identified elevated risks prior to mitigations, with Chemical and Biological capabilities assessed as likely reaching the "high risk" category under the Advanced AI Scaling Framework before safeguards were applied. We have implemented a multi-layered set of mitigations that address the identified risks, and Muse Spark demonstrates state-of-the-art refusal across a range of benchmarks related to hazardous workflows in chemistry and biology. We therefore release Muse Spark as the underlying model of Meta AI.

cs.CY↗

arXiv2026

$p1$: Better Prompt Optimization with Fewer Prompts

Prompt optimization improves language models without updating their weights by searching for a better system prompt, but its effectiveness varies widely across tasks. We study what makes a task amenable to prompt optimization. We show that the reward variance across different system prompts can be decomposed into two components: variance among responses, which captures generation stochasticity, and variance among system prompts, which captures differences in system prompt quality. Prompt optimization succeeds when variance among system prompts is sufficiently large, but fails when variance among responses dominates the variance of the system prompts. Surprisingly, we further show that scaling to more user prompts can hurt optimization by reducing variance among system prompts, especially on heterogeneous datasets where different user prompts favor different system prompts. Motivated by this insight, we propose $p1$, a simple user prompt filtering method that selects a small subset of user prompts with high variance across candidate system prompts. This subset of user prompts allows one to distinguish a good system prompt from a bad one, making system optimization easier. Experiments on reasoning benchmarks show that $p1$ substantially improves prompt optimization over training on the full dataset and outperforms strong baselines such as GEPA. Notably, training on only two prompts from AIME 24 yields a system prompt that generalizes well to other reasoning benchmarks.

cs.LG↗

arXiv2026

EDRP: Enhanced Dynamic Relay Point Protocol for Data Dissemination in Multi-hop Wireless IoT Networks

Emerging IoT applications are transitioning from battery-powered to grid-powered nodes. DRP, a contention-based data dissemination protocol, was developed for these applications. Traditional contention-based protocols resolve collisions through control packet exchanges, significantly reducing goodput. DRP mitigates this issue by employing a distributed delay timer mechanism that assigns transmission-start delays based on the average link quality between a sender and its children, prioritizing highly connected nodes for early transmission. However, our in-field experiments reveal that DRP is unable to accommodate real-world link quality fluctuations, leading to overlapping transmissions from multiple senders. This overlap triggers CSMA's random back-off delays, ultimately degrading the goodput performance. To address these shortcomings, we first conduct a theoretical analysis that characterizes the design requirements induced by real-world link quality fluctuations and DRP's passive acknowledgments. Guided by this analysis, we design EDRP, which integrates two novel components: (i) Link-Quality Aware CSMA (LQ-CSMA) and (ii) a Machine Learning-based Block Size Selection (ML-BSS) algorithm for rateless codes. LQ-CSMA dynamically restricts the back-off delay range based on real-time link quality estimates, ensuring that nodes with stronger connectivity experience shorter delays. ML-BSS algorithm predicts future link quality conditions and optimally adjusts the block size for rateless coding, reducing overhead and enhancing goodput. In-field evaluations of EDRP demonstrate an average goodput improvement of 39.43\% than the competing protocols.

cs.NI↗

arXiv2026

Architectural Insights for Post-Tornado Damage Recognition

Rapid and accurate building damage assessment in the immediate aftermath of tornadoes is critical for coordinating life-saving search and rescue operations, optimizing emergency resource allocation, and accelerating community recovery. However, current automated methods struggle with the unique visual complexity of tornado-induced wreckage, primarily due to severe domain shift from standard pre-training datasets and extreme class imbalance in real-world disaster data. To address these challenges, we introduce a systematic experimental framework evaluating 79 open-source deep learning models, encompassing both Convolutional Neural Networks (CNNs) and Vision Transformers, across over 2,300 controlled experiments on our newly curated Quad-State Tornado Damage (QSTD) benchmark dataset. Our findings reveal that achieving operational-grade performance hinges on a complex interaction between architecture and optimization, rather than architectural selection alone. Most strikingly, we demonstrate that optimizer choice can be more consequential than architecture: switching from Adam to SGD provided dramatic F1 gains of +25 to +38 points for Vision Transformer and Swin Transformer families, fundamentally reversing their ranking from bottom-tier to competitive with top-performing CNNs. Furthermore, a low learning rate of 1x10^(-4) proved universally critical, boosting average F1 performance by +10.2 points across all architectures. Our champion model, ConvNeXt-Base trained with these optimized settings, demonstrated strong cross-event generalization on the held-out Tuscaloosa-Moore Tornado Damage (TMTD) dataset, achieving 46.4% Macro F1 (+34.6 points over baseline) and retaining 85.5% Ordinal Top-1 Accuracy despite temporal and sensor domain shifts.

cs.CV↗

arXiv2025

Simple Agents Outperform Experts in Biomedical Imaging Workflow Optimization

Adapting production-level computer vision tools to bespoke scientific datasets is a critical "last mile" bottleneck. Current solutions are impractical: fine-tuning requires large annotated datasets scientists often lack, while manual code adaptation costs scientists weeks to months of effort. We consider using AI agents to automate this manual coding, and focus on the open question of optimal agent design for this targeted task. We introduce a systematic evaluation framework for agentic code optimization and use it to study three production-level biomedical imaging pipelines. We demonstrate that a simple agent framework consistently generates adaptation code that outperforms human-expert solutions. Our analysis reveals that common, complex agent architectures are not universally beneficial, leading to a practical roadmap for agent design. We open source our framework and validate our approach by deploying agent-generated functions into a production pipeline, demonstrating a clear pathway for real-world impact.

cs.CV↗

arXiv2025

Macroeconomic Forecasting and Machine Learning

We forecast the full conditional distribution of macroeconomic outcomes by systematically integrating three key principles: using high-dimensional data with appropriate regularization, adopting rigorous out-of-sample validation procedures, and incorporating nonlinearities. By exploiting the rich information embedded in a large set of macroeconomic and financial predictors, we produce accurate predictions of the entire profile of macroeconomic risk in real time. Our findings show that regularization via shrinkage is essential to control model complexity, while introducing nonlinearities yields limited improvements in predictive accuracy. Out-of-sample validation plays a critical role in selecting model architecture and preventing overfitting.

econ.EM↗

arXiv2025

Forecasting Liquidity Withdraw with Machine Learning Models

Liquidity withdrawal is a critical indicator of market fragility. In this project, I test a framework for forecasting liquidity withdrawal at the individual-stock level, ranging from less liquid stocks to highly liquid large-cap tickers, and evaluate the relative performance of competing model classes in predicting short-horizon order book stress. We introduce the Liquidity Withdrawal Index (LWI) -- defined as the ratio of order cancellations to the sum of standing depth and new additions at the best quotes -- as a bounded, interpretable measure of transient liquidity removal. Using Nasdaq market-by-order (MBO) data, we compare a spectrum of approaches: linear benchmarks (AR, HAR), and non-linear tree ensembles (XGBoost), across horizons ranging from 250\,ms to 5\,s. Beyond predictive accuracy, our results provide insights into order placement and cancellation dynamics, identify regimes where linear versus non-linear signals dominate, and highlight how early-warning indicators of liquidity withdrawal can inform both market surveillance and execution.

q-fin.RM↗

arXiv2025

An AI system to help scientists write expert-level empirical software

The cycle of scientific discovery is frequently bottlenecked by the slow, manual creation of software to support computational experiments\cite{hannay2009how}. To address this, we present Empirical Research Assistance (ERA), an AI system that creates expert-level scientific software whose goal is to maximize a quality metric. The system uses a Large Language Model (LLM) and Tree Search (TS)\cite{silver2016mastering} to systematically improve the quality metric and intelligently navigate the large space of possible solutions. ERA achieves expert-level results when it explores and integrates complex research ideas from external sources. The effectiveness of tree search is demonstrated across a diverse range of tasks. In bioinformatics, ERA discovered 40 novel methods for single-cell data analysis that outperformed the top human-developed methods on a public leaderboard. In epidemiology, ERA generated 14 models that outperformed the CDC ensemble and all other individual models for forecasting COVID-19 hospitalizations. ERA also produced expert-level software for geospatial analysis, neural activity prediction in zebrafish, and numerical solution of integrals, and a novel rule-based construction for time series forecasting. By devising and implementing novel solutions to diverse tasks, ERA represents a significant step towards accelerating scientific progress.

cs.AI↗

arXiv2025

Generative AI for Industrial Contour Detection: A Language-Guided Vision System

Industrial computer vision systems often struggle with noise, material variability, and uncontrolled imaging conditions, limiting the effectiveness of classical edge detectors and handcrafted pipelines. In this work, we present a language-guided generative vision system for remnant contour detection in manufacturing, designed to achieve CAD-level precision. The system is organized into three stages: data acquisition and preprocessing, contour generation using a conditional GAN, and multimodal contour refinement through vision-language modeling, where standardized prompts are crafted in a human-in-the-loop process and applied through image-text guided synthesis. On proprietary FabTrack datasets, the proposed system improved contour fidelity, enhancing edge continuity and geometric alignment while reducing manual tracing. For the refinement stage, we benchmarked several vision-language models, including Google's Gemini 2.0 Flash, OpenAI's GPT-image-1 integrated within a VLM-guided workflow, and open-source baselines. Under standardized conditions, GPT-image-1 consistently outperformed Gemini 2.0 Flash in both structural accuracy and perceptual quality. These findings demonstrate the promise of VLM-guided generative workflows for advancing industrial computer vision beyond the limitations of classical pipelines.

cs.CV↗

arXiv2025

Optical Counterparts to X-ray sources in LSST DP1

We present a crossmatch between a combined catalog of X-ray sources and the Vera C. Rubin Observatory Data Preview 1 (DP1) to identify optical counterparts. The six fields targeted as part of DP1 include the Extended Chandra Deep Field South (E-CDF-S), the Euclid Deep Field South (EDF-S), the Fornax Dwarf Spheroidal Galaxy (Fornax dSph), 47 Tucanae (47 Tuc) and science validation fields with low galactic and ecliptic latitude (SV\_95\_-25 and SV\_38\_7, respectively). We find matches to 2314 of 3830 X-ray sources. We also compare our crossmatch to DP1 in the E-CDF-S field to previous efforts to identify optical counterparts. The probability of a chance coincidence match varies across each DP1 field, with overall high reliability in the E-CDF-S field, and lower proportion of high-reliability matches in the other fields. The majority of previously known sources that we detect are, unsurprisingly, active galaxies. We plot the X-ray-to-optical flux ratio against optical magnitude and color in an effort to identify Galactic accreting compact objects using a {\em Gaia} color threshold transformed to LSST $g$--$i$, but do not find any strong candidates in these primarily extragalactic counterparts. The DP1 dataset contains high-cadence photometry collected over a number of nights. We calculate the Stetson $ J $ variability index for each object under the hypothesis that X-ray counterparts tend to exhibit higher optical variability; however, the evidence is inconclusive whether our sample is more variable over DP1 timescales when compared to field objects.

astro-ph.HE↗

arXiv2025

Crowded Field Photometry with Rubin: Exploring 47 Tucanae with Data Preview 1

We analyze imaging from Data Preview 1 of the Vera C. Rubin Observatory to explore the performance of early LSST pipelines in the 47 Tucanae field. The coadd-\texttt{object} catalog demonstrates the depth and precision possible with Rubin, recovering well-defined color magnitude diagrams for 47 Tuc Small Magellanic Cloud. Unfortunately, the existing pipelines fail to recover sources within $\sim$28 pc of the cluster center, due to the extreme source density. Using Rubin's forced photometry on stars identified via Difference Imaging, we can recover sources down to $\sim$14 pc from the cluster center, and find 14744 potential cluster members with this extended dataset. While this forced photometry has significant systematics, our analysis showcases the potential for detailed structural studies of crowded fields with the Rubin Observatory.

astro-ph.GA↗

arXiv2025

47 Tuc in Rubin Data Preview 1: Exploring Early LSST Data and Science Potential

We present analyses of the early data from Rubin Observatory's Data Preview 1 (DP1) for the globular cluster 47 Tuc field. The DP1 dataset for 47 Tuc includes four nights of observations from the Rubin Commissioning Camera (LSSTComCam), covering multiple bands ($ugriy$). We address challenges of crowding in the inner region of the cluster and toward the SMC in DP1, and demonstrate improved star-galaxy separation by fitting fifth-degree polynomials to the stellar loci in color-color diagrams and applying multi-dimensional sigma clipping. We compile a catalog of 3,576 probable 47 Tuc member stars selected via a combination of isochrone, Gaia proper-motion, and color-color space matched filtering. We explore the sources of photometric scatter in the 47 Tuc color-color sequence, evaluating contributions from various potential sources, including differential extinction within the cluster. Finally, of the 72 well-characterized variables in the field, we recover five known variable stars, including three RR Lyrae and two eclipsing binaries, in the coadd-based object catalog, and identify 62 in the difference image-based object catalog. Although the DP1 lightcurves have sparse temporal sampling, they appear to follow the patterns of densely-sampled literature lightcurves well. Despite some data limitations for crowded-field stellar analysis, DP1 demonstrates the promising scientific potential for future LSST data releases.

astro-ph.SR↗

arXiv2025

D-LiFT: Improving LLM-based Decompiler Backend via Code Quality-driven Fine-tuning

As one of the key tools in many security tasks, decompilers reconstruct human-readable source code from binaries. Yet, despite recent advances, their outputs often suffer from syntactic and semantic errors and remain difficult to read. Recently, with the advent of large language models (LLMs), researchers began to explore the potential of LLMs to refine decompiler output. Nevertheless, our study of these approaches reveals their problems, such as introducing new errors and relying on unreliable accuracy validation. In this paper, we present D-LIFT, an enhanced decompiler-LLM pipeline with a fine-tuned LLM using code quality-aware reinforcement learning. Unlike prior work that overlooks preserving accuracy, D-LIFT adheres to a key principle for enhancing the quality of decompiled code: preserving accuracy while improving readability. Central to D-LIFT, we propose D-Score, an integrated code quality assessment system to score the decompiled source code from multiple aspects, and use it to guide reinforcement learning fine-tuning and to select the best output during inference. In line with our principle, D-Score assigns low scores to any inaccurate output and only awards higher scores for readability to code that passes the accuracy check. Our implementation, based on Ghidra and a range of LLMs, demonstrates significant improvements for the accurate decompiled code from the coreutils and util-linux projects. Compared to baseline LLMs without D-Score-driven fine-tuning, our trained LLMs produce 55.3% more improved decompiled functions, as measured by D-Score. Overall, D-LIFT improves the quality of 68.2% of all the functions produced by the native decompiler.

cs.CR↗

arXiv2025

Advancing Waterfall Plots for Cancer Treatment Response Assessment through Adjustment of Incomplete Follow-Up Time

Waterfall plots are a key tool in early phase oncology clinical studies for visualizing individual patients' tumor size changes and provide efficacy assessment. However, comparing waterfall plots from ongoing studies with limited follow-up to those from completed studies with long follow-up is challenging due to underestimation of tumor response in ongoing patients. To address this, we propose a novel adjustment method that projects the waterfall plot of an ongoing study to approximate its appearance with sufficient follow-up. Recognizing that waterfall plots are simply rotated survival functions of best tumor size reduction from the baseline (in percentage), we frame the problem in a survival analysis context and adjust weight of each ongoing patients in an interim look Kaplan-Meier curve by leveraging the probability of potential tumor response improvement (i.e., "censoring"). The probability of improvement is quantified through an incomplete multinomial model to estimate the best tumor size change occurrence at each scan time. The adjusted waterfall plots of experimental treatments from ongoing studies are suitable for comparison with historical controls from completed studies, without requiring individual-level data of those controls. A real-data example demonstrates the utility of this method for robust efficacy evaluations.

stat.AP↗

arXiv2025

Growing Through Experience: Scaling Episodic Grounding in Language Models

Language models (LMs) require robust episodic grounding-the capacity to learn from and apply past experiences-to excel at physical planning tasks. Current episodic grounding approaches struggle with scalability and integration, limiting their effectiveness, especially for medium-sized LMs (7B parameters). While larger LMs (70-405B parameters) possess superior hierarchical representations and extensive pre-trained knowledge, they encounter a fundamental scale paradox: despite their advanced abstraction capabilities, they lack efficient mechanisms to leverage experience streams. We propose a scalable weak-to-strong episodic learning framework that effectively transfers episodic behaviors from smaller to larger LMs. This framework integrates Monte Carlo tree search for structured experience collection with a novel distillation method, preserving the inherent LM capabilities while embedding episodic memory. Experiments demonstrate our method surpasses state-of-the-art proprietary LMs by 3.45% across diverse planning and question-answering tasks. Layer-wise probing further indicates significant improvements in task alignment, especially within deeper LM layers, highlighting stable generalization even for previously unseen scenarios with increased planning complexity-conditions where baseline methods degrade markedly.

cs.CL↗

arXiv2025

TuneComp: Joint Fine-tuning and Compression for Large Foundation Models

To reduce model size during post-training, compression methods, including knowledge distillation, low-rank approximation, and pruning, are often applied after fine-tuning the model. However, sequential fine-tuning and compression sacrifices performance, while creating a larger than necessary model as an intermediate step. In this work, we aim to reduce this gap, by directly constructing a smaller model while guided by the downstream task. We propose to jointly fine-tune and compress the model by gradually distilling it to a pruned low-rank structure. Experiments demonstrate that joint fine-tuning and compression significantly outperforms other sequential compression methods.

cs.LG↗