arXiv subjects

Qi Dong

Publications and source records attributed to Qi Dong.

At least 19 recordsLinked to original sources

arXiv2026

Stateful Evidence-Driven Retrieval-Augmented Generation with Iterative Reasoning

Retrieval-Augmented Generation (RAG) grounds Large Language Models (LLMs) in external knowledge but often suffers from flat context representations and stateless retrieval, leading to unstable performance. We propose Stateful Evidence-Driven RAG with Iterative Reasoning, a framework that models question answering as a progressive evidence accumulation process. Retrieved documents are converted into structured reasoning units with explicit relevance and confidence signals and maintained in a persistent evidence pool capturing both supportive and non-supportive information. The framework performs evidence-driven deficiency analysis to identify gaps and conflicts and iteratively refines queries to guide subsequent retrieval. This iterative reasoning process enables stable evidence aggregation and improves robustness to noisy retrieval. Experiments on multiple question answering benchmarks demonstrate consistent improvements over standard RAG and multi-step baselines, while effectively accumulating high-quality evidence and maintaining stable performance under substantial retrieval noise.

cs.CL↗

arXiv2026

Short-term electricity load forecasting with multi-frequency reconstruction diffusion

Diffusion models have emerged as a powerful method in various applications. However, their application to Short-Term Electricity Load Forecasting (STELF) -- a typical scenario in energy systems -- remains largely unexplored. Considering the nonlinear and fluctuating characteristics of the load data, effectively utilizing the powerful modeling capabilities of diffusion models to enhance STELF accuracy remains a challenge. This paper proposes a novel diffusion model with multi-frequency reconstruction for STELF, referred to as the Multi-Frequency-Reconstruction-based Diffusion (MFRD) model. The MFRD model achieves accurate load forecasting through four key steps: (1) The original data is combined with the decomposed multi-frequency modes to form a new data representation; (2) The diffusion model adds noise to the new data, effectively reducing and weakening the noise in the original data; (3) The reverse process adopts a denoising network that combines Long Short-Term Memory (LSTM) and Transformer to enhance noise removal; and (4) The inference process generates the final predictions based on the trained denoising network. To validate the effectiveness of the MFRD model, we conducted experiments on two data platforms: Australian Energy Market Operator (AEMO) and Independent System Operator of New England (ISO-NE). The experimental results show that our model consistently outperforms the compared models.

cs.LG↗

arXiv2025

Hierarchical Alignment: Surgical Fine-Tuning via Functional Layer Specialization in Large Language Models

Existing alignment techniques for Large Language Models (LLMs), such as Direct Preference Optimization (DPO), typically treat the model as a monolithic entity, applying uniform optimization pressure across all layers. This approach overlooks the functional specialization within the Transformer architecture, where different layers are known to handle distinct tasks from syntax to abstract reasoning. In this paper, we challenge this one-size-fits-all paradigm by introducing Hierarchical Alignment, a novel method that applies targeted DPO to distinct functional blocks of a model's layers: local (syntax), intermediate (logic), and global (factuality). Through a series of controlled experiments on state-of-the-art models like Llama-3.1-8B and Qwen1.5-7B using LoRA for surgical fine-tuning, our results, evaluated by a powerful LLM-as-Judge, demonstrate significant and predictable improvements. Specifically, aligning the local layers (Local-Align) enhances grammatical fluency. More importantly, aligning the global layers (Global-Align) not only improves factual consistency as hypothesized but also proves to be the most effective strategy for enhancing logical coherence, outperforming all baselines. Critically, all hierarchical strategies successfully avoid the "alignment tax" observed in standard DPO, where gains in fluency come at the cost of degraded logical reasoning. These findings establish a more resource-efficient, controllable, and interpretable path for model alignment, highlighting the immense potential of shifting from monolithic optimization to structure-aware surgical fine-tuning to build more advanced and reliable LLMs.

cs.CL↗

arXiv2025

Pre/Absence: Prompting Cultural Awareness and Understanding for Lost Architectural Heritage in Virtual Reality

Lost architectural heritage presents interpretive challenges due to vanished structures and fragmented historical records. Using Hanyuan Hall of the Tang dynasty's Daming Palace as a case study, we conducted a formative investigation with archaeologists, heritage administrators, and visitors to identify key issues in current interpretation practices. We found that these practices often compress complex cultural layers into factual summaries and rely on linear narratives that overlook the continuing reinterpretations following a site's disappearance. In response, we designed Pre/Absence, a virtual reality experience grounded in the presence-absence dialectic to interweave tangible and vanished aspects of heritage within a spatiotemporal narrative. A mixed-method study with 28 participants compared Pre/Absence to a paper-based experience. Both improved users' factual understanding, but the VR experience more strongly enhanced cultural awareness, evoked emotional engagement with loss, and encouraged critical reflection on the evolving social and political meanings of heritage. The findings suggest that VR can move beyond static reconstruction to engage users as co-constructors of cultural meaning, providing a nuanced framework for critical heritage narrative design in human-computer interaction.

cs.HC↗

arXiv2025

LacAIDes: Generative AI-Supported Creative Interactive Circuits Crafting to Enliven Traditional Lacquerware

Lacquerware, a representative craft of Chinese intangible cultural heritage, is renowned for its layered aesthetics and durability but faces declining engagement. While prior human-computer interaction research has explored embedding interactive circuits to transform lacquerware into responsive artifacts, most studies have focused on fabrication techniques rather than supporting makers in creatively designing such interactions at a low threshold. To address this gap, we present LacAIDes, a Generative AI powered creativity-support tool built on a multi-agent workflow aligned with the double diamond model of design thinking. LacAIDes enables exploration and creation of culturally grounded interactive circuits without requiring prior technical expertise. We evaluated LacAIDes in a longitudinal workshop with 34 participants using a mixed-method approach. Results show that LacAIDes demonstrated high usability, enhanced creative engagement in craft making, and encouraged critical reflection on the role of Generative AI in digital craft practices. This work contributes to human-computer interaction by introducing a novel creativity-support tool and providing empirical insights into revitalizing traditional craft making through Generative AI.

cs.HC↗

arXiv2025

HF-VTON: High-Fidelity Virtual Try-On via Consistent Geometric and Semantic Alignment

Virtual try-on technology has become increasingly important in the fashion and retail industries, enabling the generation of high-fidelity garment images that adapt seamlessly to target human models. While existing methods have achieved notable progress, they still face significant challenges in maintaining consistency across different poses. Specifically, geometric distortions lead to a lack of spatial consistency, mismatches in garment structure and texture across poses result in semantic inconsistency, and the loss or distortion of fine-grained details diminishes visual fidelity. To address these challenges, we propose HF-VTON, a novel framework that ensures high-fidelity virtual try-on performance across diverse poses. HF-VTON consists of three key modules: (1) the Appearance-Preserving Warp Alignment Module (APWAM), which aligns garments to human poses, addressing geometric deformations and ensuring spatial consistency; (2) the Semantic Representation and Comprehension Module (SRCM), which captures fine-grained garment attributes and multi-pose data to enhance semantic representation, maintaining structural, textural, and pattern consistency; and (3) the Multimodal Prior-Guided Appearance Generation Module (MPAGM), which integrates multimodal features and prior knowledge from pre-trained models to optimize appearance generation, ensuring both semantic and geometric consistency. Additionally, to overcome data limitations in existing benchmarks, we introduce the SAMP-VTONS dataset, featuring multi-pose pairs and rich textual annotations for a more comprehensive evaluation. Experimental results demonstrate that HF-VTON outperforms state-of-the-art methods on both VITON-HD and SAMP-VTONS, excelling in visual fidelity, semantic consistency, and detail preservation.

cs.CV↗

arXiv2025

Multi-Scale Manifold Alignment for Interpreting Large Language Models: A Unified Information-Geometric Framework

We present Multi-Scale Manifold Alignment(MSMA), an information-geometric framework that decomposes LLM representations into local, intermediate, and global manifolds and learns cross-scale mappings that preserve geometry and information. Across GPT-2, BERT, RoBERTa, and T5, we observe consistent hierarchical patterns and find that MSMA improves alignment metrics under multiple estimators (e.g., relative KL reduction and MI gains with statistical significance across seeds). Controlled interventions at different scales yield distinct and architecture-dependent effects on lexical diversity, sentence structure, and discourse coherence. While our theoretical analysis relies on idealized assumptions, the empirical results suggest that multi-objective alignment offers a practical lens for analyzing cross-scale information flow and guiding representation-level control.

cs.CL↗

arXiv2025

Latent Trajectory Dynamics in Large Language Models: A Manifold Evolution Framework with Empirical Validation

Understanding how latent representations evolve during generation is a central open problem in large language model interpretability. We introduce \textbf{Dynamical Manifold Evolution Theory} (DMET), a phenomenological framework that models LLM generation as a controlled dynamical system evolving along a trajectory on a low-dimensional semantic manifold. DMET formalizes the structural correspondence between Transformer components and a first-order ODE governed by a semantic potential $V$, and characterizes trajectory geometry through three falsifiable proxy metrics: state continuity $C$, attractor clustering quality $Q$, and topological persistence $P$, targeting local smoothness, meso-scale basin structure, and global topological organization, respectively. Across six model architectures, four task types, and 1,080 experimental runs, all three metrics consistently predict text quality outcomes -- log-perplexity, grammaticality, and cross-sentence coherence -- after controlling for decoding parameters, with associations surviving Benjamini--Hochberg correction. Ablation and sanity-check experiments confirm that the effects arise from genuine trajectory structure rather than static distributional artefacts. Furthermore, online monitoring of $C$ drives an adaptive decoding controller that reduces perplexity from 48.5 to 14.6 relative to a fixed-parameter baseline, demonstrating that latent dynamics characterization translates directly into actionable generation control.

cs.CL↗

arXiv2025

Emergent Hierarchical Structure in Large Language Models: An Information-Theoretic Framework for Multi-Scale Representation

Why do language models from different architecture families respond so differently to the same perturbation? We argue that the answer is not scale, but \emph{how architecture shapes information compression}. Analyzing eight Transformer models (7B--70B parameters) from the Llama and Qwen families, we show that every model spontaneously develops discrete functional boundaries dividing its layers into Local, Intermediate, and Global processing segments -- yet boundary locations and per-segment brittleness are determined overwhelmingly by architecture family rather than model size or training configuration. We formalize this regularity as the \textbf{Multi-Scale Probabilistic Generation Theory} (MSPGT), which models an autoregressive Transformer as a Hierarchical Variational Information Bottleneck system and derives a tiered set of falsifiable predictions. Three predictions are strongly confirmed: all eight models exhibit two prominent phase-transition boundaries (P1.1); Llama boundary positions are stable across a $10{\times}$ parameter range ($\mathrm{CV}{=}0.067$--$0.095$) while Qwen positions vary widely ($\mathrm{CV}{=}0.465$--$0.726$), precisely matching our strong- and weak-dominance conditions; and cross-architecture local-segment brittleness spans \textbf{three orders of magnitude} ($493{\times}$ ratio) -- a gap that architecture family alone predicts and that dwarfs any within-family or scale-driven variation.

cs.CL↗

arXiv2024

Revenue vs. Welfare: A Comprehensive Analysis of Strategic Trade-offs in Online Food Delivery Systems

This paper investigates the trade-off between short-term revenue generation and long-term social welfare optimization in online food delivery platforms. We first develop a static model that captures the equilibrium interactions among restaurants, consumers, and delivery workers, using Gross Merchandise Value (GMV) as a proxy for immediate performance. Building on this, we extend our analysis to a dynamic model that integrates evolving state variables,such as platform reputation and participant retention-to capture long-term behavior. By applying dynamic programming techniques, we derive optimal strategies that balance GMV maximization with social welfare enhancement. Extensive multi-agent simulations validate our theoretical predictions, demonstrating that while a GMV-focused approach yields strong initial gains, it ultimately undermines long-term stability. In contrast, a social welfare-oriented strategy produces more sustainable and robust outcomes. Our findings provide actionable insights for platform operators and policymakers seeking to harmonize rapid growth with long-term

cs.GT↗

arXiv2024

Exploring the Head Effect in Live Streaming Platforms: A Two-Sided Market and Welfare Analysis

We develop a comprehensive theoretical framework to analyze live streaming platforms as two-sided markets, focusing on the head effect where a small subset of elite streamers disproportionately attracts viewer attention. By constructing both static and dynamic models, we capture the interplay between network effects, content quality investments, and platform policies-such as commission structures and traffic allocation algorithms-that drive traffic concentration. Our welfare analysis demonstrates that although short-term consumer utility may benefit from concentrated viewership, long-term content diversity and overall social welfare are adversely impacted. Extensive simulations further validate our models and show that targeted policy interventions can rebalance viewer distribution and mitigate winner-takes-all dynamics. These findings offer actionable insights for platform designers and regulators in the digital economy.

cs.CY↗

arXiv2024

Short-Term Electricity-Load Forecasting by Deep Learning: A Comprehensive Survey

Short-Term Electricity-Load Forecasting (STELF) refers to the prediction of the immediate demand (in the next few hours to several days) for the power system. Various external factors, such as weather changes and the emergence of new electricity consumption scenarios, can impact electricity demand, causing load data to fluctuate and become non-linear, which increases the complexity and difficulty of STELF. In the past decade, deep learning has been applied to STELF, modeling and predicting electricity demand with high accuracy, and contributing significantly to the development of STELF. This paper provides a comprehensive survey on deep-learning-based STELF over the past ten years. It examines the entire forecasting process, including data pre-processing, feature extraction, deep-learning modeling and optimization, and results evaluation. This paper also identifies some research challenges and potential research directions to be further investigated in future work.

cs.LG↗

arXiv2024

Unveiling LLM Mechanisms Through Neural ODEs and Control Theory

This paper proposes a framework combining Neural Ordinary Differential Equations (Neural ODEs) and robust control theory to enhance the interpretability and control of large language models (LLMs). By utilizing Neural ODEs to model the dynamic evolution of input-output relationships and introducing control mechanisms to optimize output quality, we demonstrate the effectiveness of this approach across multiple question-answer datasets. Experimental results show that the integration of Neural ODEs and control theory significantly improves output consistency and model interpretability, advancing the development of explainable AI technologies.

cs.LG↗

arXiv2024

Non-autoregressive Sequence-to-Sequence Vision-Language Models

Sequence-to-sequence vision-language models are showing promise, but their applicability is limited by their inference latency due to their autoregressive way of generating predictions. We propose a parallel decoding sequence-to-sequence vision-language model, trained with a Query-CTC loss, that marginalizes over multiple inference paths in the decoder. This allows us to model the joint distribution of tokens, rather than restricting to conditional distribution as in an autoregressive model. The resulting model, NARVL, achieves performance on-par with its state-of-the-art autoregressive counterpart, but is faster at inference time, reducing from the linear complexity associated with the sequential generation of tokens to a paradigm of constant time joint inference.

cs.CV↗

arXiv2023

DocFormerv2: Local Features for Document Understanding

We propose DocFormerv2, a multi-modal transformer for Visual Document Understanding (VDU). The VDU domain entails understanding documents (beyond mere OCR predictions) e.g., extracting information from a form, VQA for documents and other tasks. VDU is challenging as it needs a model to make sense of multiple modalities (visual, language and spatial) to make a prediction. Our approach, termed DocFormerv2 is an encoder-decoder transformer which takes as input - vision, language and spatial features. DocFormerv2 is pre-trained with unsupervised tasks employed asymmetrically i.e., two novel document tasks on encoder and one on the auto-regressive decoder. The unsupervised tasks have been carefully designed to ensure that the pre-training encourages local-feature alignment between multiple modalities. DocFormerv2 when evaluated on nine datasets shows state-of-the-art performance over strong baselines e.g. TabFact (4.3%), InfoVQA (1.4%), FUNSD (1%). Furthermore, to show generalization capabilities, on three VQA tasks involving scene-text, Doc- Formerv2 outperforms previous comparably-sized models and even does better than much larger models (such as GIT2, PaLi and Flamingo) on some tasks. Extensive ablations show that due to its pre-training, DocFormerv2 understands multiple modalities better than prior-art in VDU.

cs.CV↗

arXiv2021

Visual Relationship Detection Using Part-and-Sum Transformers with Composite Queries

Computer vision applications such as visual relationship detection and human object interaction can be formulated as a composite (structured) set detection problem in which both the parts (subject, object, and predicate) and the sum (triplet as a whole) are to be detected in a hierarchical fashion. In this paper, we present a new approach, denoted Part-and-Sum detection Transformer (PST), to perform end-to-end visual composite set detection. Different from existing Transformers in which queries are at a single level, we simultaneously model the joint part and sum hypotheses/interactions with composite queries and attention modules. We explicitly incorporate sum queries to enable better modeling of the part-and-sum relations that are absent in the standard Transformers. Our approach also uses novel tensor-based part queries and vector-based sum queries, and models their joint interaction. We report experiments on two vision tasks, visual relationship detection and human object interaction and demonstrate that PST achieves state of the art results among single-stage models, while nearly matching the results of custom designed two-stage models.

cs.CV↗

arXiv2021

Adaptive dynamic programming-based adaptive-gain sliding mode tracking control for fixed-wing UAV with disturbances

This paper proposes an adaptive dynamic programming-based adaptive-gain sliding mode control (ADP-ASMC) scheme for a fixed-wing unmanned aerial vehicle (UAV) with matched and unmatched disturbances. Starting from the dynamic of fixed-wing UAV, the control-oriented model composed of attitude subsystem and airspeed subsystem is established. According to the different issues in two subsystems, two novel adaptive-gain generalized super-twisting (AGST) algorithms are developed to eliminate the effects of disturbances in two subsystems and make the system trajectories tend to the designed integral sliding manifolds (ISMs) in finite time. Then, based on the expected equivalent sliding-mode dynamics, the modified adaptive dynamic programming (ADP) approach with actor-critic (AC) structure is utilized to generate the nearly optimal control laws and achieve the nearly optimal performance of the sliding-mode dynamics. Furthermore, through the Lyapunov stability theorem, the tracking errors and the weight estimation errors of two neural networks (NNs) are all uniformly ultimately bounded (UUB). Finally, comparative simulations demonstrate the superior performance of the proposed control scheme for the fixed-wing UAV.

eess.SY↗

arXiv2021

Efficient Processing of k-regret Minimization Queries with Theoretical Guarantees

Assisting end users to identify desired results from a large dataset is an important problem for multi-criteria decision making. To address this problem, top-k and skyline queries have been widely adopted, but they both have inherent drawbacks, i.e., the user either has to provide a specific utility function or faces many results. The k-regret minimization query is proposed, which integrates the merits of top-k and skyline queries. Due to the NP-hardness of the problem, the k-regret minimization query is time consuming and the greedy framework is widely adopted. However, formal theoretical analysis of the greedy approaches for the quality of the returned results is still lacking. In this paper, we first fill this gap by conducting a nontrivial theoretical analysis of the approximation ratio of the returned results. To speed up query processing, a sampling-based method, StocPreGreed,, is developed to reduce the evaluation cost. In addition, a theoretical analysis of the required sample size is conducted to bound the quality of the returned results. Finally, comprehensive experiments are conducted on both real and synthetic datasets to demonstrate the efficiency and effectiveness of the proposed methods.

cs.DB↗