arXiv subjects

Jia Gao

Publications and source records attributed to Jia Gao.

15 recordsLinked to original sources

arXiv2026

History-informed Lagrangian Neural Networks

Forecasting the long-horizon evolution of mechanical systems from position-only observations is a pivotal yet difficult task, as hidden velocities and trajectory-specific physical properties must be inferred simultaneously. Although physics-guided neural networks like Lagrangian Neural Networks (LNNs) guarantee physical plausibility, they generally require complete state inputs and lack adaptability to changing system parameters. To break these limitations, we introduce History-informed Lagrangian Neural Networks (HiLNN). Grounded in the insight that temporal position sequences implicitly encode underlying dynamics, HiLNN employs a recurrent encoder to extract a latent context from history. This context not only reconstructs the unobserved initial velocity but also adaptively modulates the mass matrix, potential energy, and damping coefficients of a structured Lagrangian system. By leveraging a differentiable RK4 rollout scheme, the entire pipeline is optimized end-to-end under multi-step trajectory supervision and energy-consistency regularization. Empirical evaluations across conservative, dissipative, and heterogeneous variable-parameter systems show that HiLNN delivers superior long-term prediction accuracy and maintains precise energy profiles compared to state-of-the-art baselines. The source code is publicly available at https://github.com/yingtian22/History-informed-LNN.

cs.LG↗

arXiv2025

Floating-Body Hydrodynamic Neural Networks

Fluid-structure interaction is common in engineering and natural systems, where floating-body motion is governed by added mass, drag, and background flows. Modeling these dissipative dynamics is difficult: black-box neural models regress state derivatives with limited interpretability and unstable long-horizon predictions. We propose Floating-Body Hydrodynamic Neural Networks (FHNN), a physics-structured framework that predicts interpretable hydrodynamic parameters such as directional added masses, drag coefficients, and a streamfunction-based flow, and couples them with analytic equations of motion. This design constrains the hypothesis space, enhances interpretability, and stabilizes integration. On synthetic vortex datasets, FHNN achieves up to an order-of-magnitude lower error than Neural ODEs, recovers physically consistent flow fields. Compared with Hamiltonian and Lagrangian neural networks, FHNN more effectively handles dissipative dynamics while preserving interpretability, which bridges the gap between black-box learning and transparent system identification.

cs.LG↗

arXiv2025

A Fine-Tuning Approach for T5 Using Knowledge Graphs to Address Complex Tasks

With the development of deep learning technology, large language models have achieved remarkable results in many natural language processing tasks. However, these models still have certain limitations in handling complex reasoning tasks and understanding rich background knowledge. To solve this problem, this study proposed a T5 model fine-tuning method based on knowledge graphs, which enhances the model's reasoning ability and context understanding ability by introducing external knowledge graphs. We used the SQuAD1.1 dataset for experiments. The experimental results show that the T5 model based on knowledge graphs is significantly better than other baseline models in reasoning accuracy, context understanding, and the ability to handle complex problems. At the same time, we also explored the impact of knowledge graphs of different scales on model performance and found that as the scale of the knowledge graph increases, the performance of the model gradually improves. Especially when dealing with complex problems, the introduction of knowledge graphs greatly improves the reasoning ability of the T5 model. Ablation experiments further verify the importance of entity and relationship embedding in the model and prove that a complete knowledge graph is crucial to improving the various capabilities of the T5 model. In summary, this study provides an effective method to enhance the reasoning and understanding capabilities of large language models and provides new directions for future research.

cs.CL↗

arXiv2025

A Hybrid Model for Few-Shot Text Classification Using Transfer and Meta-Learning

With the continuous development of natural language processing (NLP) technology, text classification tasks have been widely used in multiple application fields. However, obtaining labeled data is often expensive and difficult, especially in few-shot learning scenarios. To solve this problem, this paper proposes a few-shot text classification model based on transfer learning and meta-learning. The model uses the knowledge of the pre-trained model for transfer and optimizes the model's rapid adaptability in few-sample tasks through a meta-learning mechanism. Through a series of comparative experiments and ablation experiments, we verified the effectiveness of the proposed method. The experimental results show that under the conditions of few samples and medium samples, the model based on transfer learning and meta-learning significantly outperforms traditional machine learning and deep learning methods. In addition, ablation experiments further analyzed the contribution of each component to the model performance and confirmed the key role of transfer learning and meta-learning in improving model accuracy. Finally, this paper discusses future research directions and looks forward to the potential of this method in practical applications.

cs.CL↗

arXiv2025

Dynamic Adaptation of LoRA Fine-Tuning for Efficient and Task-Specific Optimization of Large Language Models

This paper presents a novel methodology of fine-tuning for large language models-dynamic LoRA. Building from the standard Low-Rank Adaptation framework, this methodology further adds dynamic adaptation mechanisms to improve efficiency and performance. The key contribution of dynamic LoRA lies within its adaptive weight allocation mechanism coupled with an input feature-based adaptive strategy. These enhancements allow for a more precise fine-tuning process that is more tailored to specific tasks. Traditional LoRA methods use static adapter settings, not considering the different importance of model layers. In contrast, dynamic LoRA introduces a mechanism that dynamically evaluates the layer's importance during fine-tuning. This evaluation enables the reallocation of adapter parameters to fit the unique demands of each individual task, which leads to better optimization results. Another gain in flexibility arises from the consideration of the input feature distribution, which helps the model generalize better when faced with complicated and diverse datasets. The joint approach boosts not only the performance over each single task but also the generalization ability of the model. The efficiency of the dynamic LoRA was validated in experiments on benchmark datasets, such as GLUE, with surprising results. More specifically, this method achieved 88.1% accuracy with an F1-score of 87.3%. Noticeably, these improvements were made at a slight increase in computational costs: only 0.1% more resources than standard LoRA. This balance between performance and efficiency positions dynamic LoRA as a practical, scalable solution for fine-tuning LLMs, especially in resource-constrained scenarios. To take it a step further, its adaptability makes it a promising foundation for much more advanced applications, including multimodal tasks.

cs.CL↗

arXiv2025

Multi-Level Attention and Contrastive Learning for Enhanced Text Classification with an Optimized Transformer

This paper studies a text classification algorithm based on an improved Transformer to improve the performance and efficiency of the model in text classification tasks. Aiming at the shortcomings of the traditional Transformer model in capturing deep semantic relationships and optimizing computational complexity, this paper introduces a multi-level attention mechanism and a contrastive learning strategy. The multi-level attention mechanism effectively models the global semantics and local features in the text by combining global attention with local attention; the contrastive learning strategy enhances the model's ability to distinguish between different categories by constructing positive and negative sample pairs while improving the classification effect. In addition, in order to improve the training and inference efficiency of the model on large-scale text data, this paper designs a lightweight module to optimize the feature transformation process and reduce the computational cost. Experimental results on the dataset show that the improved Transformer model outperforms the comparative models such as BiLSTM, CNN, standard Transformer, and BERT in terms of classification accuracy, F1 score, and recall rate, showing stronger semantic representation ability and generalization performance. The method proposed in this paper provides a new idea for algorithm optimization in the field of text classification and has good application potential and practical value. Future work will focus on studying the performance of this model in multi-category imbalanced datasets and cross-domain tasks and explore the integration wi

cs.CL↗

arXiv2024

Feature Alignment-Based Knowledge Distillation for Efficient Compression of Large Language Models

This study proposes a knowledge distillation algorithm based on large language models and feature alignment, aiming to effectively transfer the knowledge of large pre-trained models into lightweight student models, thereby reducing computational costs while maintaining high model performance. Different from the traditional soft label distillation method, this method introduces a multi-layer feature alignment strategy to deeply align the intermediate features and attention mechanisms of the teacher model and the student model, maximally retaining the semantic expression ability and context modeling ability of the teacher model. In terms of method design, a multi-task loss function is constructed, including feature matching loss, attention alignment loss, and output distribution matching loss, to ensure multi-level information transfer through joint optimization. The experiments were comprehensively evaluated on the GLUE data set and various natural language processing tasks. The results show that the proposed model performs very close to the state-of-the-art GPT-4 model in terms of evaluation indicators such as perplexity, BLEU, ROUGE, and CER. At the same time, it far exceeds baseline models such as DeBERTa, XLNet, and GPT-3, showing significant performance improvements and computing efficiency advantages. Research results show that the feature alignment distillation strategy is an effective model compression method that can significantly reduce computational overhead and storage requirements while maintaining model capabilities. Future research can be further expanded in the directions of self-supervised learning, cross-modal feature alignment, and multi-task transfer learning to provide more flexible and efficient solutions for the deployment and optimization of deep learning models.

cs.CL↗

arXiv2024

Optimizing Large Language Models with an Enhanced LoRA Fine-Tuning Algorithm for Efficiency and Robustness in NLP Tasks

This study proposes a large language model optimization method based on the improved LoRA fine-tuning algorithm, aiming to improve the accuracy and computational efficiency of the model in natural language processing tasks. We fine-tune the large language model through a low-rank adaptation strategy, which significantly reduces the consumption of computing resources while maintaining the powerful capabilities of the pre-trained model. The experiment uses the QQP task as the evaluation scenario. The results show that the improved LoRA algorithm shows significant improvements in accuracy, F1 score, and MCC compared with traditional models such as BERT, Roberta, T5, and GPT-4. In particular, in terms of F1 score and MCC, our model shows stronger robustness and discrimination ability, which proves the potential of the improved LoRA algorithm in fine-tuning large-scale pre-trained models. In addition, this paper also discusses the application prospects of the improved LoRA algorithm in other natural language processing tasks, emphasizing its advantages in multi-task learning and scenarios with limited computing resources. Future research can further optimize the LoRA fine-tuning strategy and expand its application in larger-scale pre-trained models to improve the generalization ability and task adaptability of the model.

cs.CL↗

arXiv2024

Adaptive Optimization for Enhanced Efficiency in Large-Scale Language Model Training

With the rapid development of natural language processing technology, large-scale language models (LLM) have achieved remarkable results in a variety of tasks. However, how to effectively train these huge models and improve their performance and computational efficiency remains an important challenge. This paper proposes an improved method based on adaptive optimization algorithm, aiming to improve the training efficiency and final performance of LLM. Through comparative experiments on the SQuAD and GLUE data sets, the experimental results show that compared with traditional optimization algorithms (such as SGD, Momentum, AdaGrad, RMSProp and Adam), the adaptive optimization algorithm we proposed has better accuracy and F1 score. Both have achieved significant improvements, especially showed stronger training capabilities when processed large-scale texts and complex tasks. The research results verify the advantages of adaptive optimization algorithms in large-scale language model training and provide new ideas and directions for future optimization methods.

cs.AI↗

arXiv2024

Graph Neural Network-Based Entity Extraction and Relationship Reasoning in Complex Knowledge Graphs

This study proposed a knowledge graph entity extraction and relationship reasoning algorithm based on a graph neural network, using a graph convolutional network and graph attention network to model the complex structure in the knowledge graph. By building an end-to-end joint model, this paper achieves efficient recognition and reasoning of entities and relationships. In the experiment, this paper compared the model with a variety of deep learning algorithms and verified its superiority through indicators such as AUC, recall rate, precision rate, and F1 value. The experimental results show that the model proposed in this paper performs well in all indicators, especially in complex knowledge graphs, it has stronger generalization ability and stability. This provides strong support for further research on knowledge graphs and also demonstrates the application potential of graph neural networks in entity extraction and relationship reasoning.

cs.CL↗

arXiv2024

Existence theorems for a generalized Chern-Simons equation on finite graphs

Denote by $G=(V,E)$ a finite graph. We study a generalized Chern-Simons equation $$ Δu=λ\mathrm{e}^u(\mathrm{e}^{bu}-1)+4π\sum\limits_{j=1}^{N}δ_{p_j} $$ on $G$, where $λ$ and $b$ are positive constants; $N$ is a positive integer; $p_1, p_2, \cdot\cdot\cdot, p_N$ are distinct vertices of $V$ and $δ_{p_j}$ is the Dirac delta mass at $p_j$. We prove that there exists a critical value $λ_c$ such that the equation has a solution if $λ\geq λ_c$ and the equation has no solution if $λ<λ_c$. We also prove that if $λ>λ_c$ the equation has at least two solutions which include a local minimizer for the corresponding functional and a mountain-pass type solution.

math.AP↗

arXiv2021

A new distance based on minimal absent words and applications to biological sequences

A minimal absent word of a sequence x, is a sequence yt hat is not a factorof x, but all of its proper factors are factors of x as well. The set of minimal absent words uniquely defines the sequence itself. In recent times minimal absent words have been used in order to compare sequences. In fact, to do this, one can compare the sets of their minimal absent words. Chairungasee and Crochemorein [2] define a distance between pairs of sequences x and y, where the symmetric difference of the sets of minimal absent words of x and y is involved. Here, weconsider a different distance, introduced in [1], based on a specific subset of such symmetric difference that, in our opinion, better capture the different features ofthe considered sequences. We show the result of some experiments where the distance is tested on a dataset of genetic sequences by 11 living species, in order to compare the new distance with the ones existing in literature.

cs.FL↗

arXiv2019

On the cyclic regularities of strings

Regularities in strings are often related to periods and covers, which have extensively been studied, and algorithms for their efficient computation have broad application. In this paper we concentrate on computing cyclic regularities of strings, in particular, we propose several efficient algorithms for computing: (i) cyclic periodicity; (ii) all cyclic periodicity; (iii) maximal local cyclic periodicity; (iv) cyclic covers.

cs.DS↗

arXiv2017

Optimal Computation of Overabundant Words

The observed frequency of the longest proper prefix, the longest proper suffix, and the longest infix of a word $w$ in a given sequence $x$ can be used for classifying $w$ as avoided or overabundant. The definitions used for the expectation and deviation of $w$ in this statistical model were described and biologically justified by Brendel et al. (J Biomol Struct Dyn 1986). We have very recently introduced a time-optimal algorithm for computing all avoided words of a given sequence over an integer alphabet (Algorithms Mol Biol 2017). In this article, we extend this study by presenting an $\mathcal{O}(n)$-time and $\mathcal{O}(n)$-space algorithm for computing all overabundant words in a sequence $x$ of length $n$ over an integer alphabet. Our main result is based on a new non-trivial combinatorial property of the suffix tree $\mathcal{T}$ of $x$: the number of distinct factors of $x$ whose longest infix is the label of an explicit node of $\mathcal{T}$ is no more than $3n-4$. We further show that the presented algorithm is time-optimal by proving that $\mathcal{O}(n)$ is a tight upper bound for the number of overabundant words. Finally, we present experimental results, using both synthetic and real data, which justify the effectiveness and efficiency of our approach in practical terms.

cs.DS↗

arXiv2016

Optimal Computation of Avoided Words

The deviation of the observed frequency of a word $w$ from its expected frequency in a given sequence $x$ is used to determine whether or not the word is avoided. This concept is particularly useful in DNA linguistic analysis. The value of the standard deviation of $w$, denoted by $std(w)$, effectively characterises the extent of a word by its edge contrast in the context in which it occurs. A word $w$ of length $k>2$ is a $ρ$-avoided word in $x$ if $std(w) \leq ρ$, for a given threshold $ρ< 0$. Notice that such a word may be completely absent from $x$. Hence computing all such words na\"ıvely can be a very time-consuming procedure, in particular for large $k$. In this article, we propose an $O(n)$-time and $O(n)$-space algorithm to compute all $ρ$-avoided words of length $k$ in a given sequence $x$ of length $n$ over a fixed-sized alphabet. We also present a time-optimal $O(σn)$-time and $O(σn)$-space algorithm to compute all $ρ$-avoided words (of any length) in a sequence of length $n$ over an alphabet of size $σ$. Furthermore, we provide a tight asymptotic upper bound for the number of $ρ$-avoided words and the expected length of the longest one. We make available an open-source implementation of our algorithm. Experimental results, using both real and synthetic data, show the efficiency of our implementation.

cs.DS↗