arXiv subjects

Thang Nguyen

Publications and source records attributed to Thang Nguyen.

At least 19 recordsLinked to original sources

arXiv2026

In-Context Inpainting for Time Series Forecasting

We propose ICI-Time, a novel framework that reframes time series forecasting as a visual inpainting task, leveraging the generalisation power of large vision models (LVMs). Unlike methods that require specialised temporal architectures and extensive domain-specific training, ICI-Time transforms time series into structured visual representations (area charts) and applies visual in-context learning, reformulating forecasting as pattern completion within a grid-structured prompt that pre-trained vision transformers can solve without fine-tuning or architectural modification. Temporal dependencies are represented through spatial layout, with a consistent, invertible mapping between numerical and visual domains. Extensive experiments across epidemiology, meteorology, and power systems demonstrate that ICI-Time performs competitively against deep learning baselines and shows promising adaptability under limited-data settings, introducing a new paradigm that bridges temporal and visual domains.

cs.AI↗

arXiv2026

Stability for boundary actions of cocompact lattices in Euclidean buildings

When $X$ is a locally compact Euclidean building, the isometry group of $X$ acts by homeomorphisms on the space of $k$-simplices in the visual boundary of $X$. We consider perturbations of these actions for discrete groups of isometries acting with compact quotient on $X$, showing that all small enough perturbations are semi-conjugate to the original action. This proves in particular that, when $Q$ is any parabolic subgroup in a semisimple $p$-adic Lie group $G$, the induced action of a cocompact lattice in $G$ has a topologically stable action on $G/Q$.

math.DS↗

arXiv2025

MA-RAG: Multi-Agent Retrieval-Augmented Generation via Collaborative Chain-of-Thought Reasoning

We present MA-RAG, a Multi-Agent framework for Retrieval-Augmented Generation (RAG) that addresses the inherent ambiguities and reasoning challenges in complex information-seeking tasks. Unlike conventional RAG methods that rely on end-to-end fine-tuning or isolated component enhancements, MA-RAG orchestrates a collaborative set of specialized AI agents: Planner, Step Definer, Extractor, and QA Agents, each responsible for a distinct stage of the RAG pipeline. By decomposing tasks into subtasks such as query disambiguation, evidence extraction, and answer synthesis, and enabling agents to communicate intermediate reasoning via chain-of-thought prompting, MA-RAG progressively refines retrieval and synthesis while maintaining modular interpretability. Extensive experiments on multi-hop and ambiguous QA benchmarks, including NQ, HotpotQA, 2WikimQA, and TriviaQA, demonstrate that MA-RAG significantly outperforms standalone LLMs and existing RAG methods across all model scales. Notably, even a small LLaMA3-8B model equipped with MA-RAG surpasses larger standalone LLMs, while larger variants (LLaMA3-70B and GPT-4o-mini) set new state-of-the-art results on challenging multi-hop datasets. Ablation studies reveal that both the planner and extractor agents are critical for multi-hop reasoning, and that high-capacity models are especially important for the QA agent to synthesize answers effectively. Beyond general-domain QA, MA-RAG generalizes to specialized domains such as medical QA, achieving competitive performance against domain-specific models without any domain-specific fine-tuning. Our results highlight the effectiveness of collaborative, modular reasoning in retrieval-augmented systems: MA-RAG not only improves answer accuracy and robustness but also provides interpretable intermediate reasoning steps, establishing a new paradigm for efficient and reliable multi-agent RAG.

cs.CL↗

arXiv2024

SemiKong: Curating, Training, and Evaluating A Semiconductor Industry-Specific Large Language Model

Large Language Models (LLMs) have demonstrated the potential to address some issues within the semiconductor industry. However, they are often general-purpose models that lack the specialized knowledge needed to tackle the unique challenges of this sector, such as the intricate physics and chemistry of semiconductor devices and processes. SemiKong, the first industry-specific LLM for the semiconductor domain, provides a foundation that can be used to develop tailored proprietary models. With SemiKong 1.0, we aim to develop a foundational model capable of understanding etching problems at an expert level. Our key contributions include (a) curating a comprehensive corpus of semiconductor-related texts, (b) creating a foundational model with in-depth semiconductor knowledge, and (c) introducing a framework for integrating expert knowledge, thereby advancing the evaluation process of domain-specific AI models. Through fine-tuning a pre-trained LLM using our curated dataset, we have shown that SemiKong outperforms larger, general-purpose LLMs in various semiconductor manufacturing and design tasks. Our extensive experiments underscore the importance of developing domain-specific LLMs as a foundation for company- or tool-specific proprietary models, paving the way for further research and applications in the semiconductor domain. Code and dataset will be available at https://github.com/aitomatic/semikong

cs.CL↗

arXiv2024

MP-PINN: A Multi-Phase Physics-Informed Neural Network for Epidemic Forecasting

Forecasting temporal processes such as virus spreading in epidemics often requires more than just observed time-series data, especially at the beginning of a wave when data is limited. Traditional methods employ mechanistic models like the SIR family, which make strong assumptions about the underlying spreading process, often represented as a small set of compact differential equations. Data-driven methods such as deep neural networks make no such assumptions and can capture the generative process in more detail, but fail in long-term forecasting due to data limitations. We propose a new hybrid method called MP-PINN (Multi-Phase Physics-Informed Neural Network) to overcome the limitations of these two major approaches. MP-PINN instils the spreading mechanism into a neural network, enabling the mechanism to update in phases over time, reflecting the dynamics of the epidemics due to policy interventions. Experiments on COVID-19 waves demonstrate that MP-PINN achieves superior performance over pure data-driven or model-driven approaches for both short-term and long-term forecasting.

cs.AI↗

arXiv2024

Reward-RAG: Enhancing RAG with Reward Driven Supervision

In this paper, we introduce Reward-RAG, a novel approach designed to enhance the Retrieval-Augmented Generation (RAG) model through Reward-Driven Supervision. Unlike previous RAG methodologies, which focus on training language models (LMs) to utilize external knowledge retrieved from external sources, our method adapts retrieval information to specific domains by employing CriticGPT to train a dedicated reward model. This reward model generates synthesized datasets for fine-tuning the RAG encoder, aligning its outputs more closely with human preferences. The versatility of our approach allows it to be effectively applied across various domains through domain-specific fine-tuning. We evaluate Reward-RAG on publicly available benchmarks from multiple domains, comparing it to state-of-the-art methods. Our experimental results demonstrate significant improvements in performance, highlighting the effectiveness of Reward-RAG in improving the relevance and quality of generated responses. These findings underscore the potential of integrating reward models with RAG to achieve superior outcomes in natural language generation tasks.

cs.CL↗

arXiv2024

Rigidity of compact rank one symmetric spaces

We consider rigidity properties of compact symmetric spaces $X$ with metric $g_0$ of rank one. Suppose $g$ is another Riemannian metric on $X$ with sectional curvature $\kappa$ bounded by $0 \leq \kappa \leq 1$. If $g$ equals $g_0$ outside a convex proper subset of $X$, then $g$ is isometric with $g_0$. We also exhibit examples of surfaces showing that the nonnegativity of the curvature is needed. Our main result complements earlier results on other symmetric spaces by Gromov and Schroeder-Ziller.

math.DG↗

arXiv2024

Embedding products of trees into higher rank

We show that there exists a quasi-isometric embedding of the product of $n$ copies of $\mathbb{H}_{\mathbb{R}}^2$ into any symmetric space of non-compact type of rank $n$, and there exists a bi-Lipschitz embedding of the product of $n$ copies of the $3$-regular tree $T_3$ into any thick Euclidean building of rank $n$ with co-compact affine Weyl group. This extends a previous result of Fisher--Whyte. The proof is purely geometrical, and the result also applies to the non Bruhat--Tits buildings.

math.GR↗

arXiv2024

Spatially temporally distributed informative path planning for multi-robot systems

This paper investigates the problem of informative path planning for a mobile robotic sensor network in spatially temporally distributed mapping. The robots are able to gather noisy measurements from an area of interest during their movements to build a Gaussian Process (GP) model of a spatio-temporal field. The model is then utilized to predict the spatio-temporal phenomenon at different points of interest. To spatially and temporally navigate the group of robots so that they can optimally acquire maximal information gains while their connectivity is preserved, we propose a novel multistep prediction informative path planning optimization strategy employing our newly defined local cost functions. By using the dual decomposition method, it is feasible and practical to effectively solve the optimization problem in a distributed manner. The proposed method was validated through synthetic experiments utilizing real-world data sets.

cs.RO↗

arXiv2023

Symbolic Path-guided Test Cases for Models with Data and Time

This paper focuses on generating test cases from timed symbolic transition systems. At the heart of the generation process are symbolic execution techniques on data and time. Test cases look like finite symbolic trees with verdicts on their leaves and are based on a user-specified finite symbolic path playing the role of a test purpose. Generated test cases handle data involved in time constraints and uninitialized parameters, leveraging the advantages of symbolic execution techniques.

cs.FL↗

arXiv2023

Boundary actions of lattices and $C^0$ local semi-rigidity

We consider actions of cocompact lattices in semisimple Lie groups of the noncompact type on their boundaries $G/Q$, $Q$ a parabolic group, the so-called standard actions. We show that perturbations of the standard action in the homeomorphism group continuously factor onto the original standard action by a semi-conjugacy close to the identity. This generalizes works by Bowden, Mann, Manning and Weisman in the setting of negative curvature or Gromov hyperbolic groups. Finally, we also construct perturbations of the action of lattices on the geodesic boundary which are not $C^0$ semi-conjugate to the original action.

math.DS↗

arXiv2022

Marked length spectrum rigidity for relatively hyperbolic groups

We consider a coarse version of the marked length spectrum rigidity: given a group with two left invariant metrics, if the marked length spectrum (the translation length function) under the two metrics are the same, then the two metrics are uniformly close. We prove the rigidity theorem for relatively hyperbolic groups. This generalizes a result of Fujiwara.

math.GT↗

arXiv2021

Cheeger-Gromoll Splitting Theorem for groups

We study a notion of curvature for finitely generated groups which serves as a role of Ricci curvature for Riemannian manifolds. We prove an analog of Cheeger-Gromoll splitting theorem. As a consequence, we give a geometric characterization of virtually abelian groups. We also explore the relation between this notion of curvature and the growth of groups.

math.GR↗

arXiv2021

Carnot metrics, Dynamics and Local Rigidity

This paper develops new techniques for studying smooth dynamical systems in the presence of a \CC metric. Principally, we employ the theory of Margulis-Mostow, Métivier, Mitchell and Pansu on tangent cones to establish resonances between Lyapunov exponents. We apply these results in three different settings. First, we explore rigidity properties of smooth dominated splittings for Anosov diffeomorphisms and flows via associated smooth \CC metrics. Second, we obtain local rigidity properties of higher hyperbolic rank metrics in a neighborhood of a locally symmetric one. For the latter application we also prove structural stability of the Brin-Pesin asymptotic holonomy group for frame flows. Finally, we obtain local rigidity properties for uniform lattice actions on the ideal boundary of quaternionic and octonionic symmetric spaces.

math.DS↗

arXiv2019

Quasi-isometric embeddings of symmetric spaces and lattices: reducible case

We study quasi-isometric embeddings of symmetric spaces and non-uniform irreducible lattices in semisimple higher rank Lie groups. We show that any quasi-isometric embedding between symmetric spaces of the same rank can be decomposed into a product of quasi-isometric embeddings into irreducible symmetric spaces. We thus extend earlier rigidity results about quasi-isometric embeddings to the setting of semisimple Lie groups. We also present some examples when the rigidity does not hold, including first examples in which every flat is mapped into multiple flats.

math.DG↗

arXiv2019

On the Control Effort in Output Feedback Sliding Mode Control of Sampled-Data Systems

In this paper, the problem of output feedback sliding mode control of linear sampled-data multi-input multi-output systems is considered. Existing sliding mode control schemes can attenuate the influence of an external disturbance by driving system states onto a sliding surface. However, they can exhibit high gains during transients, which can be $O(1/T)$ where $T$ is the sampling time period. To address this problem, a new strategy, which employs disturbance approximation, is proposed so that the control effort will be $O(1)$. The new method avoids deadbeat phenomena and hence, it will be less sensitive to noise. Theoretical analysis is provided to show the convergence and robustness of the proposed method. Simulations were conducted to show the efficiency of the proposed approach.

eess.SY↗

arXiv2018

Vector Learning for Cross Domain Representations

Recently, generative adversarial networks have gained a lot of popularity for image generation tasks. However, such models are associated with complex learning mechanisms and demand very large relevant datasets. This work borrows concepts from image and video captioning models to form an image generative framework. The model is trained in a similar fashion as recurrent captioning model and uses the learned weights for image generation. This is done in an inverse direction, where the input is a caption and the output is an image. The vector representation of the sentence and frames are extracted from an encoder-decoder model which is initially trained on similar sentence and image pairs. Our model conditions image generation on a natural language caption. We leverage a sequence-to-sequence model to generate synthetic captions that have the same meaning for having a robust image generation. One key advantage of our method is that the traditional image captioning datasets can be used for synthetic sentence paraphrases. Results indicate that images generated through multiple captions are better at capturing the semantic meaning of the family of captions.

cs.LG↗

arXiv2018

Semantic Sentence Embeddings for Paraphrasing and Text Summarization

This paper introduces a sentence to vector encoding framework suitable for advanced natural language processing. Our latent representation is shown to encode sentences with common semantic information with similar vector representations. The vector representation is extracted from an encoder-decoder model which is trained on sentence paraphrase pairs. We demonstrate the application of the sentence representations for two different tasks -- sentence paraphrasing and paragraph summarization, making it attractive for commonly used recurrent frameworks that process text. Experimental results help gain insight how vector representations are suitable for advanced language embedding.

cs.CL↗