arXiv subjects

Thong Q. Nguyen

Publications and source records attributed to Thong Q. Nguyen.

11 recordsLinked to original sources

arXiv2026

PPT-Eval: A Benchmark for Computer-Use Agents on PowerPoint Tasks

Creating and editing slides is a rich, multimodal activity that is ubiquitous in professional and educational settings, making it an ideal testbed for real-world computer-use agents. Microsoft PowerPoint is among the most widely adopted and feature-rich environments for presentation creation. We introduce PPT-Eval, a benchmark of 120 PowerPoint tasks across 12 files that cover both content creation and presentation editing scenarios, organized by difficulty. A central challenge in this domain is evaluation: tasks are complex, multimodal, and often admit many valid solutions. Moreover, today's agents frequently make only partial progress, which binary success metrics fail to capture. To address this, we design a robust evaluation framework to help create task-specific rubrics for PowerPoint tasks, taking inspiration from and building on past works for rubric-based evaluation. These rubrics award partial credit for intermediate steps, penalize unnecessary changes and poor aesthetics, and provide natural language feedback. This nuanced approach proves highly effective, achieving a Kendall's τ-b correlation of 0.77 with human judgments. We find that existing frontier agents still struggle with solving PowerPoint tasks, with strong models like Claude-4.5-Opus achieving only a 45% success rate and an average partial score of 57%. The benchmark is located at: https://microsoft.github.io/ppteval.

cs.LG↗

arXiv2026

Office Comprehension Benchmark

We introduce Office Comprehension Bench (OCB), the first public benchmark to jointly evaluate LLM systems on Word, Excel, and PowerPoint comprehension over native file formats (.docx, .xlsx, .pptx) and their variants. OCB consists of two tracks. File Fidelity Q&A tests structural and visual perception of office artifacts - tables, charts, embedded images, formulas, and app-specific elements such as headers, speaker notes, and named ranges. Domain Q&A tests expert-level reasoning grounded in real-world industry documents across 12 professional domains, with queries requiring multi-step analysis and synthesis across documents. Each reference answer is decomposed into atomic, binary-gradable claims, and an ensemble of LLM judges scores responses against each claim independently. Even the strongest frontier system in its default reasoning mode reaches only about 59.3% on Domain Q&A; increasing thinking depth within a tier does not move performance materially, while moving to a higher product tier yields modest gains. We release the dataset, evaluation tooling, judge prompt, and a public leaderboard.

cs.CL↗

arXiv2025

VerificAgent: Domain-Specific Memory Verification for Scalable Oversight of Aligned Computer-Use Agents

Continual memory augmentation lets computer-using agents (CUAs) learn from prior interactions, but unvetted memories can encode domain-inappropriate or unsafe heuristics--spurious rules that drift from user intent and safety constraints. We introduce VerificAgent, a scalable oversight framework that treats persistent memory as an explicit alignment surface. VerificAgent combines (1) an expert-curated seed of domain knowledge, (2) iterative, trajectory-based memory growth during training, and (3) a post-hoc human fact-checking pass to sanitize accumulated memories before deployment. Evaluated on OSWorld productivity tasks and additional adversarial stress tests, VerificAgent improves task reliability, reduces hallucination-induced failures, and preserves interpretable, auditable guidance--without additional model fine-tuning. By letting humans correct high-impact errors once, the verified memory acts as a frozen safety contract that future agent actions must satisfy. Our results suggest that domain-scoped, human-verified memory offers a scalable oversight mechanism for CUAs, complementing broader alignment strategies by limiting silent policy drift and anchoring agent behavior to the norms and safety constraints of the target domain.

cs.LG↗

arXiv2023

Natural Language Commanding via Program Synthesis

We present Semantic Interpreter, a natural language-friendly AI system for productivity software such as Microsoft Office that leverages large language models (LLMs) to execute user intent across application features. While LLMs are excellent at understanding user intent expressed as natural language, they are not sufficient for fulfilling application-specific user intent that requires more than text-to-text transformations. We therefore introduce the Office Domain Specific Language (ODSL), a concise, high-level language specialized for performing actions in and interacting with entities in Office applications. Semantic Interpreter leverages an Analysis-Retrieval prompt construction method with LLMs for program synthesis, translating natural language user utterances to ODSL programs that can be transpiled to application APIs and then executed. We focus our discussion primarily on a research exploration for Microsoft PowerPoint.

cs.LG↗

arXiv2021

Autoencoders on FPGAs for real-time, unsupervised new physics detection at 40 MHz at the Large Hadron Collider

In this paper, we show how to adapt and deploy anomaly detection algorithms based on deep autoencoders, for the unsupervised detection of new physics signatures in the extremely challenging environment of a real-time event selection system at the Large Hadron Collider (LHC). We demonstrate that new physics signatures can be enhanced by three orders of magnitude, while staying within the strict latency and resource constraints of a typical LHC event filtering system. This would allow for collecting datasets potentially enriched with high-purity contributions from new physics processes. Through per-layer, highly parallel implementations of network layers, support for autoencoder-specific losses on FPGAs and latent space based inference, we demonstrate that anomaly detection can be performed in as little as $80\,$ns using less than 3% of the logic resources in the Xilinx Virtex VU9P FPGA. Opening the way to real-life applications of this idea during the next data-taking campaign of the LHC.

physics.ins-det↗

arXiv2020

Data Augmentation at the LHC through Analysis-specific Fast Simulation with Deep Learning

We present a fast simulation application based on a Deep Neural Network, designed to create large analysis-specific datasets. Taking as an example the generation of W+jet events produced in sqrt(s)= 13 TeV proton-proton collisions, we train a neural network to model detector resolution effects as a transfer function acting on an analysis-specific set of relevant features, computed at generation level, i.e., in absence of detector effects. Based on this model, we propose a novel fast-simulation workflow that starts from a large amount of generator-level events to deliver large analysis-specific samples. The adoption of this approach would result in about an order-of-magnitude reduction in computing and storage requirements for the collision simulation workflow. This strategy could help the high energy physics community to face the computing challenges of the future High-Luminosity LHC.

physics.comp-ph↗

arXiv2020

Adversarially Learned Anomaly Detection on CMS Open Data: re-discovering the top quark

We apply an Adversarially Learned Anomaly Detection (ALAD) algorithm to the problem of detecting new physics processes in proton-proton collisions at the Large Hadron Collider. Anomaly detection based on ALAD matches performances reached by Variational Autoencoders, with a substantial improvement in some cases. Training the ALAD algorithm on 4.4 fb-1 of 8 TeV CMS Open Data, we show how a data-driven anomaly detection and characterization would work in real life, re-discovering the top quark by identifying the main features of the t-tbar experimental signature at the LHC.

hep-ex↗

arXiv2020

Interaction networks for the identification of boosted $H\to b\overline{b}$ decays

We develop an algorithm based on an interaction network to identify high-transverse-momentum Higgs bosons decaying to bottom quark-antiquark pairs and distinguish them from ordinary jets that reflect the configurations of quarks and gluons at short distances. The algorithm's inputs are features of the reconstructed charged particles in a jet and the secondary vertices associated with them. Describing the jet shower as a combination of particle-to-particle and particle-to-vertex interactions, the model is trained to learn a jet representation on which the classification problem is optimized. The algorithm is trained on simulated samples of realistic LHC collisions, released by the CMS Collaboration on the CERN Open Data Portal. The interaction network achieves a drastic improvement in the identification performance with respect to state-of-the-art algorithms.

hep-ex↗

arXiv2020

JEDI-net: a jet identification algorithm based on interaction networks

We investigate the performance of a jet identification algorithm based on interaction networks (JEDI-net) to identify all-hadronic decays of high-momentum heavy particles produced at the LHC and distinguish them from ordinary jets originating from the hadronization of quarks and gluons. The jet dynamics are described as a set of one-to-one interactions between the jet constituents. Based on a representation learned from these interactions, the jet is associated to one of the considered categories. Unlike other architectures, the JEDI-net models achieve their performance without special handling of the sparse input jet representation, extensive pre-processing, particle ordering, or specific assumptions regarding the underlying detector geometry. The presented models give better results with less model parameters, offering interesting prospects for LHC applications.

hep-ex↗

arXiv2019

Topology classification with deep learning to improve real-time event selection at the LHC

We show how event topology classification based on deep learning could be used to improve the purity of data samples selected in real time at at the Large Hadron Collider. We consider different data representations, on which different kinds of multi-class classifiers are trained. Both raw data and high-level features are utilized. In the considered examples, a filter based on the classifier's score can be trained to retain ~99% of the interesting events and reduce the false-positive rate by as much as one order of magnitude for certain background processes. By operating such a filter as part of the online event selection infrastructure of the LHC experiments, one could benefit from a more flexible and inclusive selection strategy while reducing the amount of downstream resources wasted in processing false positives. The saved resources could be translated into a reduction of the detector operation cost or into an effective increase of storage and processing capabilities, which could be reinvested to extend the physics reach of the LHC experiments.

hep-ex↗

arXiv2019

Variational Autoencoders for New Physics Mining at the Large Hadron Collider

Using variational autoencoders trained on known physics processes, we develop a one-sided threshold test to isolate previously unseen processes as outlier events. Since the autoencoder training does not depend on any specific new physics signature, the proposed procedure doesn't make specific assumptions on the nature of new physics. An event selection based on this algorithm would be complementary to classic LHC searches, typically based on model-dependent hypothesis testing. Such an algorithm would deliver a list of anomalous events, that the experimental collaborations could further scrutinize and even release as a catalog, similarly to what is typically done in other scientific domains. Event topologies repeating in this dataset could inspire new-physics model building and new experimental searches. Running in the trigger system of the LHC experiments, such an application could identify anomalous events that would be otherwise lost, extending the scientific reach of the LHC.

hep-ex↗