检索结果-内蒙古大学图书馆

SMT(LIA) Sampling with High Diversity

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Lai, Yong Li, Junjie Luo, Chuan College of Computer Science and Technology Jilin University Changchun130012 China Key Laboratory of Symbolic Computation and Knowledge Engineering Ministry of Education Jilin University Changchun130012 China School of Software Beihang University Beijing100191 China

Satisfiability Modulo Linear Integer Arithmetic, SMT(LIA) for short, is pivotal across various critical domains. Previous research has primarily focused on SMT solving techniques. However, in practical applications such as software and hardware testing, there is a need to generate a diverse set of solutions for use as test inputs. We have developed the first sampling framework that integrates local search with CDCL(T) techniques, named HighDiv, capable of generating a highly diverse set of solutions for constraints under linear integer theory. Initially, in the local search phase, we introduced a novel operator called boundary-aware movement. This operator performs random moves by considering the current state’s constraints on variables, thereby enhancing the diversity of variables during the search process. Furthermore, we have conducted an in-depth study of the preprocessing and variable initialization mechanisms within the framework, which significantly enhances the efficiency of subsequent local searches. Lastly, we use the solutions obtained from local search sampling as additional constraints to further explore the solution space using the stochastic CDCL(T) method. Experimental results demonstrate that HighDiv generates solutions with greater diversity compared to the state-of-the-art SMT(LIA) sampling tool, MeGASampler. Copyright © 2025, The Authors. All rights reserved.

关键词： Software testing

PNESR-DDI: An Effective Drug-Drug Interaction Prediction Model Based on Pretraining Method and Enhanced Subgraph Reconstruction

学校读者我要写书评

暂无评论

PNESR-DDI: An Effective Drug-Drug Interaction Prediction Mod...

IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

作者： Ke Chen Xiaosong Han Xiaoran Li Yanchun Liang Dong Xu Renchu Guan Key Laboratory for Symbol Computation and Knowledge Engineering of National Education Ministry College of Software Jilin University Changchun China Key Laboratory for Symbol Computation and Knowledge Engineering of National Education Ministry College of Computer Science and Technology Jilin University Changchun China Zhuhai Laboratory of Key Laboratory of Symbol Computation and Knowledge Engineering of Ministry of Education School of Computer Science Zhuhai College of Science and Technology Zhuhai China Department of Electrical Engineering and Computer Science Christopher S. Bond Life Sciences Center University of Missouri Columbia USA

ISBN: (数字)9798350386226

ISBN: (纸本)9798350386233

Drug-Drug Interaction (DDI) task plays a crucial role in clinical treatment and drug development. Recently, deep learning methods have been successfully applied for DDI prediction. However, training deep learning models always need large amount of data, while known DDIs are scarce. To address this challenge, a graph neural network-based DDI prediction model named PNESR-DDI is proposed, which compensates for the lack of DDIs by enriching drug representations. First, to obtain initial node representations that incorporate rich semantic information from the biomedical knowledge graph (KG), a link prediction pre-training method on external KG is proposed in the node embedding pre-training module. Then, considering the large scale of the KG, subgraph extraction for the target drug pairs is introduced to reduce noise and decrease computational complexity in the subgraph anchoring module. After that, the subgraph is updated, and node similarities are propagated in the subgraph reconstruction module. Based on the node similarity scores, the subgraph is pruned and reconstructed, which adjusts node representations to be more conducive to DDI prediction. Finally, the drug embeddings, subgraph representations, and drug fingerprint features are concatenated to predict DDIs. PNESRDDI is evaluated on two benchmark DDI datasets: DrugBank and TWOSIDES. Experiment results show that PNESR-DDI achieves better performance than baselines. Ablation results validate the effectiveness of the pre-training method and the adaptive subgraph reconstruction strategy.

关键词： Drugs Deep learning Adaptation models Visualization Biological system modeling Noise knowledge graphs Predictive models Fingerprint recognition Transformers

Decision Mamba: Reinforcement Learning via Hybrid Selective Sequence Modeling

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Huang, Sili Hu, Jifeng Yang, Zhejian Yang, Liwei Luo, Tao Chen, Hechang Sun, Lichao Yang, Bo Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education China School of Artificial Intelligence Jilin University China Institute of High Performance Computing A*STAR Singapore Lehigh University BethlehemPA United States

Recent works have shown the remarkable superiority of transformer models in reinforcement learning (RL), where the decision-making problem is formulated as sequential generation. Transformer-based agents could emerge with self-improvement in online environments by providing task contexts, such as multiple trajectories, called in-context RL. However, due to the quadratic computation complexity of attention in transformers, current in-context RL methods suffer from huge computational costs as the task horizon increases. In contrast, the Mamba model is renowned for its efficient ability to process long-term dependencies, which provides an opportunity for in-context RL to solve tasks that require long-term memory. To this end, we first implement Decision Mamba (DM) by replacing the backbone of Decision Transformer (DT). Then, we propose a Decision Mamba-Hybrid (DM-H) with the merits of transformers and Mamba in high-quality prediction and long-term memory. Specifically, DM-H first generates high-value sub-goals from long-term memory through the Mamba model. Then, we use sub-goals to prompt the transformer, establishing high-quality predictions. Experimental results demonstrate that DM-H achieves state-of-the-art in long and short-term tasks, such as D4RL, Grid World, and Tmaze benchmarks. Regarding efficiency, the online testing of DM-H in the long-term task is 28× times faster than the transformer-based baselines. Copyright © 2024, The Authors. All rights reserved.

关键词： Reinforcement learning

PRG: Prompt-Based Distillation Without Annotation via Proxy Relational Graph

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Xu, Yijin Liu, Jialun Wei, Hualiang Li, Wenhui The College of Computer Science and Technology The Key Laboratory of Symbolic Computation and Knowledge Engineer Ministry of Education Jilin University Jilin Changchun130012 China Baidu VIS. Beijing100000 China

In this paper, we propose a new distillation method for extracting knowledge from Large Foundation Models (LFM) into lightweight models, introducing a novel supervision mode that does not require manually annotated data. While LFMs exhibit exceptional zero-shot classification abilities across datasets, relying solely on LFM-generated embeddings for distillation poses two main challenges: LFM’s task-irrelevant knowledge and the high density of features. The transfer of task-irrelevant knowledge could compromise the student model’s discriminative capabilities, and the high density of features within target domains obstructs the extraction of discriminative knowledge essential for the task. To address this issue, we introduce the Proxy Relational Graph (PRG) method. We initially extract task-relevant knowledge from LFMs by calculating a weighted average of logits obtained through text prompt embeddings. Then we construct sample-class proxy graphs for LFM and student models, respectively, to model the correlation between samples and class proxies. Then, we achieve the distillation of selective knowledge by aligning the relational graphs produced by both the LFM and the student model. Specifically, the distillation from LFM to the student model is achieved through two types of alignment: 1) aligning the sample nodes produced by the student model with those produced by the LFM, and 2) aligning the edge relationships in the student model’s graph with those in the LFM’s graph. Our experimental results validate the effectiveness of PRG, demonstrating its ability to leverage the extensive knowledge base of LFMs while skillfully circumventing their inherent limitations in focused learning scenarios. Notably, in our annotation-free framework, PRG achieves an accuracy of 76.23% (T: 77.9%) on CIFAR-100 and 72.44% (T: 75.3%) on the ImageNet-1K. Copyright © 2024, The Authors. All rights reserved.

关键词： knowledge graph

Learning Stable Task-Level Manifold for Few-Shot Learning

学校读者我要写书评

暂无评论

Learning Stable Task-Level Manifold for Few-Shot Learning

Proceedings of the Digital Image Computing: Technqiues and Applications (DICTA)

作者： Siyi Yu Wei Luo Gang Li Bo Yang College of Computer Science and Technology Jilin University China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education China School of Information Technology Deakin University Geelong Australia

Few-shot learning (FSL) aims to learn to new concepts based on very limited data. One of the main challenges in FSL is the use of pretrained embeddings whose dimension is too high for the small sample size. While the large dimensionality helps capture diverse features in the training example, not all the features are relevant to the task at hand. Such task-irrelevant features from pre-trained embeddings can negatively impact the performance of the FSL model. To address this challenge, we propose a simple yet effective way to learn task-level manifolds that explicitly eliminate the negative effect of nuisance features. One key novelty of our approach is to address class imbalance in the training data by injecting supervised information through key virtual samples, which we named Manifold Support Points. Empirically, we demonstrate that the stability of the learnt task-level manifolds and their effectiveness in challenging tasks in FSL. Our approach shows competitive performance on several benchmark datasets, including tieredImageNet and CUB.

关键词：

Towards More Accurate and Complete Iris Segmentation Using Hybrid Transformer U-Net

学校读者我要写书评

暂无评论

Towards More Accurate and Complete Iris Segmentation Using H...

IEEE International Joint Conference on Biometrics (IJCB)

作者： Ye Sun Yinan Lu Yuanning Liu Xiaodong Zhu College of Computer Science and Technology Jilin University Changchun China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education

ISBN: (纸本)9781665463959

Iris images captured in less-constrained environments often suffer from adverse noise, challenging many existing segmentation algorithms. In this paper, we propose an efficient Hybrid Transformer U-Net (HTU-Net) to address this dilemma. Unlike previous studies that only focus on utilizing popular CNN technology to predict iris masks accurately, HTU-Net can simultaneously obtain segmentation masks and parameterized pupillary and limbic boundaries by a multi-task network, further enabling CNN-based iris segmentation to be applied in any regular iris recognition systems. We explore the application of the Transformer in iris segmentation and propose a hybrid encoder that employs convolutional layers to extract local intensity features and the Transformer to capture long-range associative information. For decoding, we adopt a novel Multi-Head Dilated Attention to exploit the multi-scale contextual information by gating mechanism, thus emphasizing the important features and rendering powerful representations. Inspired by the consistent class characteristics of iris, we further devise a Pyramid Center-Aware Module to capture the global structural context of iris from a categorical perspective to improve performance. Experimental results show that our method, with fewer parameters than previous approaches, achieves competitive or new state-of-the-art performance in both iris segmentation and localization on three challenging iris datasets. Code will be released at https://***/Syloveslife/HTU-Net.

关键词： Image segmentation Biological system modeling Transformers Feature extraction Rendering (computer graphics) Prediction algorithms Multitasking

UAV Deployment Optimization in UAV-assisted Wireless Communications

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhang, Xueqi Wang, Aimin Sun, Geng Liu, Lingling Zhang, Jing College of Computer Science and Technology Jilin University Changchun130012 China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun130012 China College of Communication Engineering Jilin University Changchun130012 China

Due to the fact that the locations of base stations (BSs) cannot be changed after they are installed, it is very difficult to communicate directly with remote user equipment (UE), which will directly affect the lifespan of the system. Unmanned aerial vehicles (UAVs) offer a hopeful solution as mobile relays for fifth-generation wireless communications due to the flexible and cost-effective deployment. However, with the limited onboard energy of UAV and slow progress in energy storage technology, it is a key challenge to achieve the energy-efficient communication. Therefore, in this article, we study a wireless communication network using a UAV as a high-altitude relay, and formulate a UAV relay deployment optimization problem (URDOP) to minimize the energy consumption of system by optimizing the deployment of UAV, including the locations and number of UAV hover points. Since the formulated URDOP is a mixed-integer programming problem, it presents a significant challenge for conventional gradient-based approaches. To this end, we propose a self-adaptive differential evolution with a variable population size (SaDEVPS) algorithm to solve the formulated URDOP. The performance of proposed SaDEVPS is verified through simulations, and the results show that it can successfully decrease the energy consumption of system when compared to other benchmark algorithms across multiple instances. Copyright © 2024, The Authors. All rights reserved.

关键词： Unmanned aerial vehicles (UAV)

Unsupervised Sentence Representation Learning with Frequency-induced Adversarial Tuning and Incomplete Sentence Filtering

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Wang, Bing Li, Ximing Yang, Zhiyao Guan, Yuanyuan Li, Jiayin Wang, Shengsheng College of Computer Science and Technology Jilin University China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University China School of Humanities Jilin University China

Pre-trained Language Model (PLM) is nowadays the mainstay of Unsupervised Sentence Representation Learning (USRL). However, PLMs are sensitive to the frequency information of words from their pre-training corpora, resulting in anisotropic embedding space, where the embeddings of high-frequency words are clustered but those of low-frequency words disperse sparsely. This anisotropic phenomenon results in two problems of similarity bias and information bias, lowering the quality of sentence embeddings. To solve the problems, we fine-tune PLMs by leveraging the frequency information of words and propose a novel USRL framework, namely Sentence Representation Learning with Frequency-induced Adversarial tuning and Incomplete sentence filtering (SLT-FAI). We calculate the word frequencies over the pre-training corpora of PLMs and assign words thresholding frequency labels. With them, (1) we incorporate a similarity discriminator used to distinguish the embeddings of high-frequency and low-frequency words, and adversarially tune the PLM with it, enabling to achieve uniformly frequency-invariant embedding space;and (2) we propose a novel incomplete sentence detection task, where we incorporate an information discriminator to distinguish the embeddings of original sentences and incomplete sentences by randomly masking several low-frequency words, enabling to emphasize the more informative low-frequency words. Our SLT-FAI is a flexible and plug-and-play framework, and it can be integrated with existing USRL techniques. We evaluate SLT-FAI with various backbones on benchmark datasets. Empirical results indicate that SLT-FAI can be superior to the existing USRL baselines. Our code is released in https://***/wangbing1416/SLT-FAI. Copyright © 2023, The Authors. All rights reserved.

关键词： Embeddings

Mask2Edge: Masking dependencies and dynamically capturing pixel differences in edge detection

学校读者我要写书评

暂无评论

Expert Systems with Applications 2025年 286卷

作者： Zhou, Jianhang Zhao, Hongwei Qu, Daikun Xing, Long College of Computer Science and Technology Jilin University Jilin Changchun130012 China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Jilin Changchun130012 China Army Artillery and Air Defense Academy Sergeant School Liaoning Shenyang110000 China

Edge detection plays an important role in computer vision tasks. Deep learning-based edge detectors commonly rely on encoding the long and short-term dependencies of pixel values to mine contextual information. They strongly focus on all positions in the image, ignoring the potential issue of over-encoding. Furthermore, most of these models have not attempted to leverage the inherent properties of edges. In this paper, we introduce a query-based edge detector named Mask2Edge, which is capable of masking dependencies and dynamically capturing pixel differences. Specifically, we first devise a masking strategy based on the sparsity of edges to alleviate the over-encoding issue. We propose a Region-guided Masked Attention, which adapts to edge detection and is capable of constraining cross-attention with appropriate masking intensity to extract relatively complete local features. Subsequently, we design a structure to capture the pixel differences that can help identify edges. We introduce dynamic convolutions into edge detection and refine the application scope and generation method of attention weights to effectively perceive changes in pixel gradients. Extensive experiments demonstrate the superiority of Mask2Edge compared with state-of-the-art methods. The source code is available at https://***/zhoujh2020/Mask2Edge. © 2025 Elsevier Ltd

关键词： Edge detection