检索结果-内蒙古大学图书馆

11th International Conference on Computer, Control, Informatics and its Applications, IC3INA 2024

作者： Anggara, Muchamad Galih Hindasyah, Ahmad Amda, Julfa Muhammad Waskita, A.A. Graduate Program of Informatics Engineering Universitas Pamulang Pamulang Indonesia Research Center for Computing National Research and Innovation Agency Cibinong Indonesia Research Center for Data and Information Sciences National Research and Innovation Agency Bandung Indonesia

ISBN: (纸本)9798331542313

Accurate image classification of microscopic pathogenic fungi is a challenge in the field of mycology. This study proposes a Swin Transformer-based approach to automate the identification process. The Swin Transformer model, known for its ability to capture complex spatial interactions, is adapted from ImageNet’s pre-trained model. The adaptation process is focused on fine-tuning the final classification layer to match the 5 classes of fungi present in the dataset. This classification layer is replaced by a new fully-connected layer that has the number of output according to the number of fungal classes. The weights on these layers are randomly initialized and trained in conjunction with several final convolutional layers. In addition, regularization techniques such as dropouts are applied to prevent overfitting. Adaptation using 5-fold cross-validation was carried out to find the best fold with hyperparameters including AdamW optimizer with a learning rate of 10−3, StepLR scheduler, batch size of 128 images, and a total of 9 epochs / iterations. After that, fine-tuning is carried out on the best fold with hyperparameters including AdamW optimizer with a learning rate of 10−5, StepLR scheduler, batch size of 32 images, and a total of 100 epochs / iterations. The final result showed excellent performance, with accuracy, precision, recall and an average F1-score of 95%. These results show the potential of the Swin Transformer in becoming an effective tool for mycologists. © 2024 IEEE.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Debiasing Computer Vision Models using data Augmentation based Adversarial Techniques 3

Debiasing Computer Vision Models using Data Augmentation bas...

引用

3rd Workshop on Bias, Risk, Ethical AI, Explainability and the Role of Logic and Logic Programming, BEWARE 2024

作者： Kumar, Teerath Mandal, Abhishek Leavy, Susan Little, Suzanne Mileo, Alessandra Bendechache, Malika CRT-AI School of Computing Dublin City University Dublin Ireland Insight SFI Research Center for Data Analytics School of Computing Dublin City University Dublin Ireland Insight SFI Research Center for Data Analytics School of Information and Communication Studies University College Dublin Dublin Ireland ADAPT & Lero Research Centres School of Computer Science University of Galway Galway Ireland

Deep learning models in computer vision, such as Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs), have been found to exhibit significant biases related to factors such as gender and ethnicity. These biases often originate from inherent imbalances in training data predominantly sourced from the internet. In this study, we aim to address gender bias in computer vision models by curating a specialized dataset that highlights gender-related disparities. Additionally, we measure dataset diversity across six datasets (FFHQ, WIKI, IMDB, LFW, UTK Faces, diverse dataset), five professions (CEO, engineer, nurse, politician, and teacher) and different query retrieval tasks using the Image Similarity Score (ISS). To reduce learned gender biases and increase data diversity, we propose adversarial data augmentation techniques that specifically target facial regions within images. These techniques, named Partial Mix (PM), that partially mixes two gendered faces in a squared pattern, and Noise Addition (NA), that adds noise to the facial region, are designed to mitigate bias. Our experimental results demonstrate increased data diversity across the six datasets and professions, along with reduction in gender bias for CNN-based models. However, these adversarial techniques were less effective in reducing bias for Vision Transformers. This discrepancy highlights the unique challenges for bias mitigation posed by ViTs. Consistent with prior research, our findings indicate that ViTs learn from a broader set of visual cues compared to CNNs. This increased sensitivity makes ViTs more prone to amplifying biases, emphasizing the need for tailored bias mitigation strategies when deploying these models in real-world applications. © 2024 Copyright for this paper by its authors.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

FedHKD: A Hierarchical Federated Learning Approach Integrating Clustering and Knowledge Distillation for Non-IID data 1

引用

2nd International Conference on Artificial Intelligence Security and Privacy, AIS and P 2024

作者： Hu, Shiwen Wang, Changji Li, Yuan Liu, Zhen Liu, Ning Gan, Qingqing School of Information Science and Technology Guangdong University of Foreign Studies Guangzhou China Guangdong Engineering Research Center of Data Security Governance and Privacy Computing Guangzhou China

ISBN: (数字)9789819611485

ISBN: (纸本)9789819611478

Federated learning allows decentralized model training while preserving data privacy. However, Non-IID data poses significant challenges, leading to performance degradation and increased communication overhead. This paper introduces FedHKD, a hierarchical FL algorithm that integrates client clustering and knowledge distillation to address Non-IID challenges. By clustering clients with similar data distributions and transferring knowledge across clusters, FedHKD improves model accuracy while reducing communication costs. Experimental results on Fashion-MNIST and CIFAR-10 datasets demonstrate up to 14.24% improvement in accuracy and a substantial reduction in communication overhead, proving its effectiveness in Non-IID environments. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

data Catalogs with Artificial Intelligence and Active Metadata: A Case Study of China Mobile 6

Data Catalogs with Artificial Intelligence and Active Metada...

引用

6th International Conference on Next Generation data-Driven Networks, NGDN 2024

作者： Wang, Jibin Guo, Yanmei Jiang, Yong Shang, Jing Chen, Zhuo Liu, Yu Hu, Qingyuan Yin, Zheng China Mobile Information Technology Center Beijing China Cloud Computing and Big Data Research Institute China Academy of Information and Communications Technology Beijing China

ISBN: (纸本)9798350388374

In the context of the rapid development of artificial intelligence, telecommunications operators have accumulated an increasing amount of data. China Mobile has developed a set of practical methods of enhanced data catalog in response to current issues such as difficulties in scheduling heterogeneous data from multiple sources, lack of macroscopic views of data usage, unintelligent data operation methods, and high risk of data leakage. By leveraging data virtualization, active metadata management, machine learning, along with other large-scale model technologies, China Mobile has crafted an enhanced data catalog, improving the data management efficiency and business decision-making process, further strengthening its competitiveness in the telecommunication industry. © 2024 IEEE.

关键词： Information leakage

来源：评论

学校读者我要写书评

暂无评论

Solving key challenges in collider physics with foundation models

引用

Physical Review D 2025年第5期111卷 L051504-L051504页

作者： Vinicius Mikuni Benjamin Nachman National Energy Research Scientific Computing Center Berkeley Lab Berkeley California 94720 USA Physics Division Lawrence Berkeley National Laboratory Berkeley California 94720 USA Berkeley Institute for Data Science University of California Berkeley California 94720 USA

Foundation models are neural networks that are capable of simultaneously solving many problems. Large language foundation models like ChatGPT have revolutionized many aspects of daily life, but their impact for science is not yet clear. In this paper, we use a new foundation model for hadronic jets to solve three key challenges in collider physics. In particular, we show how experiments can (1) save significant computing power when developing reconstruction algorithms, (2) perform a complete uncertainty quantification for high-dimensional measurements, and (3) search for new physics with model agnostic methods using low-level inputs. In each case, there are significant computational or methodological challenges with current methods that limit the science potential of deep learning algorithms. By solving each problem, we take jet foundation models beyond proof-of-principle studies and into the toolkit of practitioners.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Benchmarking data Heterogeneity Evaluation Approaches for Personalized Federated Learning

Benchmarking Data Heterogeneity Evaluation Approaches for P...

引用

International Workshop on Trustworthy Federated Learning, FL 2024

作者： Li, Zhilong Wu, Xiaohu Tang, Xiaoli He, Tiantian Ong, Yew-Soon Chen, Mengmeng Liu, Qiqi Lao, Qicheng Yu, Han Beijing University of Posts and Telecommunications Beijing China College of Computing and Data Science Nanyang Technological University Singapore Singapore Institute of High Performance Computing Agency for Science Technology and Research Singapore Singapore Centre for Frontier AI Research Agency for Science Technology and Research Singapore Singapore School of Engineering Westlake University Hangzhou China

ISBN: (纸本)9783031822391

There is growing research interest in measuring the statistical heterogeneity of clients’ local datasets. Such measurements are used to estimate the suitability for collaborative training of personalized federated learning (PFL) models. Currently, these research endeavors are taking place in silos and there is a lack of a unified benchmark to provide a fair and convenient comparison among various approaches in common settings. We aim to bridge this important gap in this paper. The proposed benchmarking framework currently includes six representative approaches. Extensive experiments have been conducted to compare these approaches under five standard non-IID FL settings, providing much needed insights into which approaches are advantageous under which settings. The proposed framework offers useful guidance on the suitability of various data divergence measures in FL systems. It is beneficial for keeping related research activities on the right track in terms of: (1) designing PFL schemes, (2) selecting appropriate data heterogeneity evaluation approaches for specific FL application scenarios, and (3) addressing fairness issues in collaborative model training. The code is available at https://***/Xiaoni-61/DH-Benchmark. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

Generative Deduplication For Socia Media data Selection

Generative Deduplication For Socia Media Data Selection

引用

2024 Conference on Empirical Methods in Natural Language Processing, EMNLP 2024

作者： Li, Xianming Li, Jing Department of Computing The Hong Kong Polytechnic University Hong Kong Research Centre on Data Science & Artificial Intelligence The Hong Kong Polytechnic University Hong Kong

ISBN: (纸本)9798891761681

Social media data exhibits severe redundancy caused by its noisy nature. It leads to increased training time and model bias in its processing. To address this issue, we propose a novel Generative Deduplication framework for social media data selection by removing semantically duplicate data. While related work involves data selection in task-specific training, our model acts as an efficient pre-processing method to universally enhance social media NLP pipelines. Specifically, we train a generative model via self-supervised learning to predict a keyword to capture the semantics of noisy social media text for deduplication. Meanwhile, time-dimensional Gaussian noise is added to improve training complexity and avoid learning trivial features. Extensive experiments suggest that our model can better reduce training samples while improving performance than baselines. The results show our model's potential to broadly advance social media language understanding in effectiveness and efficiency. © 2024 Association for Computational Linguistics.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Investigating the Impact of Internal Inputs on Mmultisensory Integration: A Study of External Multisensory Inputs and Internal Arousal 32

Investigating the Impact of Internal Inputs on Mmultisensory...

引用

32nd Irish Conference on Artificial Intelligence and Cognitive Science, AICS 2024

作者： Azizi, Zahra Chan, Jason Ward, Tomas Setti, Annalisa Insight SFI Research Centre for Data Analytics School of Computing Dublin City University Ireland School of Applied Psychology University College Cork Cork Ireland

Multisensory integration, the brain's ability to combine information from different sensory modalities, is influenced by both external stimuli and internal states, such as arousal, motivation, and emotion. This study explores the balance between internal and external inputs and their effect on multisensory integration, with a focus on arousal. We hypothesize that multisensory integration abilities vary depending on both internal states and external inputs. In our experiment, we tested 23 participants using the Sound-Induced Flash Illusion (external input) and arousal-cued images (internal arousal). Sensory sensitivity was also assessed using the Highly Sensitive Person (HSP) scale. Preliminary findings revealed no significant differences in response accuracy or confidence when participants were presented with cue-induced arousal versus neutral images across different age groups. However, greater sensory sensitivity was linked to enhanced internal judgment. These results suggest that while arousal-cued images may not be the most effective method for inducing internal arousal in this context, individual differences play a role in how internal states affect multisensory integration. To further explore this, we plan to implement the cold-water pressor technique, along with physiological monitoring (GSR and heart rate), to more effectively induce and control internal arousal. By exploring how internal states influence multisensory integration, our findings could inform strategies to improve sensory processing and behavioral outcomes in individuals with deficits, such as those with Parkinson's disease or age-related challenges. These interventions may enhance motor coordination, cognitive function, and quality of life while also providing insights for broader applications in mental health and technology. © 2024 Copyright for this paper by its authors.

关键词： Neurodegenerative diseases

来源：评论

学校读者我要写书评

暂无评论

TikTok Video Cluster Analysis Based on Trending Topic 9th

TikTok Video Cluster Analysis Based on Trending Topic

引用

9th International Conference on computing and Informatics, ICOCI 2023

作者： Bakar, Juhaida Abu Sharimi, Nur Azmielia Muhammad Shahril, Mohd Azrul Edzwan Azmi, Nur Syafiqah Harun, Nor Hazlyna Awang, Hapini Bakar, Nur Syafiqah Abu Data Management and Software Solution Research Lab School of Computing Universiti Utara Malaysia Kedah Sintok06010 Malaysia Data Science Research Lab School of Computing Universiti Utara Malaysia Kedah Sintok06010 Malaysia Institute for Advanced and Smart Digital Opportunities School of Computing Universiti Utara Malaysia Kedah Sintok06010 Malaysia Faculty of Medicine and Health Sciences Universiti Sains Islam Malaysia Negeri Sembilan Nilai71800 Malaysia

ISBN: (纸本)9789819995882

TikTok is a popular social networking application that offers trend research and is a valuable source for users. However, this is often misconstrued for content, which may not be suitable for children due to misappropriate content. This study aims to improve user perception of TikTok by using topic modelling and clustering techniques to identify trending topics in TikTok videos. The research uses Latent Dirichlet Allocation (LDA) and K-means clustering techniques to enhance the recognition of local and global topics across text documents. The methodology includes data collection, data pre-processing, clustering, topic modelling, and results. Ten subjects associated with trending TikTok videos are displayed using the LDA algorithm, and the generated topics are used to produce an Inter-topic Distance Map. The method’s effectiveness is evaluated using log-likelihood score and perplexity measurements. It has a log-likelihood score of 5579 and a perplexity score of 287. A good model is one with a higher log-likelihood and lower perplexity. The study ex-tracts popular TikTok topics using both the LDA topic modelling technique and the K-means clustering algorithm. © 2024, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Cluster analysis

来源：评论

学校读者我要写书评

暂无评论

BeLLM: Backward Dependency Enhanced Large Language Model for Sentence Embeddings

BeLLM: Backward Dependency Enhanced Large Language Model for...

引用

2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL 2024

作者： Li, Xianming Li, Jing Department of Computing Hong Kong Polytechnic University Hong Kong Research Centre on Data Science and Artificial Intelligence Hong Kong Polytechnic University Hong Kong

ISBN: (纸本)9798891761148

Sentence embeddings are crucial in measuring semantic similarity. Most recent studies employed large language models (LLMs) to learn sentence embeddings. Existing LLMs mainly adopted autoregressive architecture without explicit backward dependency modeling. Therefore, we examined the effects of backward dependencies in LLMs for semantic similarity measurements. Concretely, we propose a novel model: backward dependency enhanced large language model (BeLLM). It learns sentence embeddings via transforming specific attention layers from uni- to bi-directional. We extensively experiment across various semantic textual similarity (STS) tasks and downstream applications. BeLLM achieves state-of-the-art performance in varying scenarios. It shows that auto-regressive LLMs benefit from backward dependencies for sentence embeddings. © 2024 Association for Computational Linguistics.

关键词： Embeddings

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：