检索结果-内蒙古大学图书馆

Gradient-Based Differential kWTA Network With Application to Competitive Coordination of Multiple Robots

学校读者我要写书评

暂无评论

IEEE/CAA Journal of Automatica Sinica 2022年第8期9卷 1452-1463页

作者： Mei Liu Xiaoyan Zhang Mingsheng Shang Long Jin School of Information Science and Engineering Lanzhou UniversityLanzhou 730000 State Key Laboratory of Tibetan Intelligent Information Processing and Application Xining 810016China Chongqing Key Laboratory of Big Data and Intelligent Computing Chongqing Institute of Green and Intelligent TechnologyChinese Academy of SciencesChongqing 400714China Shenzhen Research Institute Lanzhou UniversityShenzhen 518063China

Aiming at the k-winners-take-all(kWTA)operation,this paper proposes a gradient-based differential kWTA(GDk WTA)*** obtaining the network,theorems and related proofs are provided to guarantee the exponential convergence and noise resistance of the proposed GD-kWTA ***,numerical simulations are conducted to substantiate the preferable performance of the proposed network as compared with the traditional ***,the GD-k WTA network,backed with a consensus filter,is utilized as a robust control scheme for modeling the competition behavior in the multi-robot coordination,thereby further demonstrating its effectiveness and feasibility.

关键词： Consensus filter k-winners-take-all(kWTA) multirobot coordination noise resistance

Great Models Think Alike: Improving Model Reliability via Inter-Model Latent Agreement

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Deng, Ailin Xiong, Miao Hooi, Bryan School of Computing National University of Singapore Singapore Institute of Data Science National University of Singapore Singapore

Reliable application of machine learning is of primary importance to the practical deployment of deep learning methods. A fundamental challenge is that models are often unreliable due to overconfidence (Hendrycks & Gimpel, 2017). In this paper, we estimate a model's reliability by measuring the agreement between its latent space, and the latent space of a foundation model. However, it is challenging to measure the agreement between two different latent spaces due to their incoherence, e.g., arbitrary rotations and different dimensionality. To overcome this incoherence issue, we design a neighborhood agreement measure between latent spaces and find that this agreement is surprisingly well-correlated with the reliability of a model's predictions. Further, we show that fusing neighborhood agreement into a model's predictive confidence in a post-hoc way significantly improves its reliability. Theoretical analysis and extensive experiments on failure detection across various datasets verify the effectiveness of our method on both in-distribution and out-of-distribution settings. Copyright © 2023, The Authors. All rights reserved.

关键词： Reliability

Leveraging Old Knowledge to Continually Learn New Classes in Medical Images

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Chee, Evelyn Lee, Mong Li Hsu, Wynne School of Computing National University of Singapore Singapore Institute of Data Science National University of Singapore Singapore

Class-incremental continual learning is a core step towards developing artificial intelligence systems that can continuously adapt to changes in the environment by learning new concepts without forgetting those previously learned. This is especially needed in the medical domain where continually learning from new incoming data is required to classify an expanded set of diseases. In this work, we focus on how old knowledge can be leveraged to learn new classes without catastrophic forgetting. We propose a framework that comprises of two main components: (1) a dynamic architecture with expanding representations to preserve previously learned features and accommodate new features;and (2) a training procedure alternating between two objectives to balance the learning of new features while maintaining the model’s performance on old classes. Experiment results on multiple medical datasets show that our solution is able to achieve superior performance over state-of-the-art baselines in terms of class accuracy and forgetting. Copyright © 2023, The Authors. All rights reserved.

关键词： Knowledge management

GraphCleaner: Detecting Mislabelled Samples in Popular Graph Learning Benchmarks

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Li, Yuwen Xiong, Miao Hooi, Bryan School of Computing National University of Singapore Singapore Institute of Data Science National University of Singapore Singapore

Label errors have been found to be prevalent in popular text, vision, and audio datasets, which heavily influence the safe development and evaluation of machine learning algorithms. Despite increasing efforts towards improving the quality of generic data types, such as images and texts, the problem of mislabel detection in graph data remains underexplored. To bridge the gap, we explore mislabelling issues in popular real-world graph datasets and propose GRAPHCLEANER, a post-hoc method to detect and correct these mislabelled nodes in graph datasets. GRAPHCLEANER combines the novel ideas of 1) Synthetic Mislabel dataset Generation, which seeks to generate realistic mislabels;and 2) Neighborhood-Aware Mislabel Detection, where neighborhood dependency is exploited in both labels and base classifier predictions. Empirical evaluations on 6 datasets and 6 experimental settings demonstrate that GRAPHCLEANER outperforms the closest baseline, with an average improvement of 0.14 in F1 score, and 0.16 in MCC. On real-data case studies, GRAPHCLEANER detects real and previously unknown mislabels in popular graph benchmarks: PubMed, Cora, CiteSeer and OGB-arxiv;we find that at least 6.91% of PubMed data is mislabelled or ambiguous, and simply removing these mislabelled data can boost evaluation performance from 86.71% to 89.11%1 © 2023, CC BY.

关键词： Benchmarking

Trajectory set Empowered Hypergraph Transformer for Mobile Sensor Based Traffic Prediction

学校读者我要写书评

暂无评论

Trajectory set Empowered Hypergraph Transformer for Mobile S...

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Hanyuan Zhang Xinyu Zhang Qize Jiang Liang Li Baihua Zheng Weiwei Sun School of Computer Science Shanghai Key Laboratory of Data Science Fudan University Shanghai Institute of Intelligent Electronics & Systems School of Computing and Information Systems Singapore Management University

Traffic speed prediction is vital for intelligent transportation systems. However, most existing methods focus on costly static sensors. In contrast, utilizing GPS devices from vehicles as mobile sensors offers a cost-effective means to gather dynamic traffic data. Despite the presence of historical trajectory data, mobile sensor-based traffic prediction remains under-explored. Existing methods often treat trajectories as substitutes for static sensors, missing the full utilization of the spatial-temporal signals within the complete trajectory set. To address this, we propose TrajHGT, a novel trajectory set empowered hypergraph transformer model that captures trafficrelated spatial-temporal features through adaptive attention and fusion mechanisms in both the trajectory hypergraph space and the road graph space. Real dataset experiments demonstrate the superiority of TrajHGT.

关键词：

From Counting Stations to City-Wide Estimates: data-Driven Bicycle Volume Extrapolation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Kaiser, Silke K. Klein, Nadja Kaack, Lynn H. Data Science Lab Hertie School Berlin10117 Germany Scientific Computing Center Karlsruhe Institute of Technology Karlsruhe76131 Germany

Shifting to cycling in urban areas reduces greenhouse gas emissions and improves public health. Street-level bicycle volume information would aid cities in planning targeted infrastructure improvements to encourage cycling and provide civil society with evidence to advocate for cyclists’ needs. Yet, the data currently available to cities and citizens often only comes from sparsely located counting stations. This paper extrapolates bicycle volume beyond these few locations to estimate bicycle volume for the entire city of Berlin. We predict daily and average annual daily street-level bicycle volumes using machine-learning techniques and various public data sources. These include app-based crowdsourced data, infrastructure, bike-sharing, motorized traffic, socioeconomic indicators, weather, and holiday data. Our analysis reveals that the best-performing model is XGBoost, and crowdsourced cycling and infrastructure data are most important for the prediction. We further simulate how collecting short-term counts at predicted locations improves performance. By providing ten days of such sample counts for each predicted location to the model, we are able to halve the error and greatly reduce the variability in performance among predicted locations. Copyright © 2024, The Authors. All rights reserved.

关键词： Bicycles

Calligraphy Alphabet Perception Using Artificial Intelligence

学校读者我要写书评

暂无评论

Calligraphy Alphabet Perception Using Artificial Intelligenc...

International Conference on science Technology Engineering and Management (ICONSTEM)

作者： Venkatesh S D. Jeevitha J. Anitha Gnanaselvi Department of Data Science and Business Systems School of Computing SRM Institute of Science and Technology Kattankulathur Chennai Department of Computer Science and Engineering Jeppiaar Engineering College Chennai

ISBN: (数字)9798350365092

ISBN: (纸本)9798350365108

The human brain has a simple time analyzing and processing images. The brain is able to rapidly deconstruct and distinguish an image's various components when the eye perceives it. With the Convolutional Neural Network (CNN) as its foundation, this research suggests deep learning conceptual models. When the algorithms are compared, it becomes clear that CNN-based classification of handwritten alphabets performs better than other algorithms in terms of accuracy. The Manual Net, Alex Net, and LeNet Architectures are among the CNN algorithms employed in this research. The convulational layer, max pooling, flattening, feature assortment, rectifier lined unit, and completely linked softmaxx layers are each components of the aforementioned designs. The proposed network is tested using an image dataset comprising 530 training photos and 2756 testing images. The top precision and cost-efficient model will be used in the Django context to build a handler line for supplying the appeal to be recognized and obtaining the productivity outcome of recognized appeal.

关键词： Training Productivity Accuracy Rectifiers User interfaces Mathematical models Classification algorithms

Towards Robust Universal Information Extraction: Benchmark, Evaluation, and Solution

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Zhu, Jizhao Shi, Akang Li, Zixuan Bai, Long Jin, Xiaolong Guo, Jiafeng Cheng, Xueqi Key Laboratory of Network Data Science and Technology Institute of Computing Technology Chinese Academy of Sciences China School of Computer Science Shenyang Aerospace University Shenyang China

In this paper, we aim to enhance the robustness of Universal Information Extraction (UIE) by introducing a new benchmark dataset, a comprehensive evaluation, and a feasible solution. Existing robust benchmark datasets have two key limitations: 1) They generate only a limited range of perturbations for a single Information Extraction (IE) task, which fails to evaluate the robustness of UIE models effectively;2) They rely on small models or handcrafted rules to generate perturbations, often resulting in unnatural adversarial examples. Considering the powerful generation capabilities of Large Language Models (LLMs), we introduce a new benchmark dataset for Robust UIE, called RUIE-Bench, which utilizes LLMs to generate more diverse and realistic perturbations across different IE tasks. Based on this dataset, we comprehensively evaluate existing UIE models and reveal that both LLM-based models and other models suffer from significant performance drops. To improve robustness and reduce training costs, we propose a data-augmentation solution that dynamically selects hard samples for iterative training based on the model’s inference loss. Experimental results show that training with only 15% of the data leads to an average 7.5% relative performance improvement across three IE tasks. Copyright © 2025, The Authors. All rights reserved.

关键词： Information retrieval

RIDE: Enhancing Large Language Model Alignment through Restyled In-Context Learning Demonstration Exemplars

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Hua, Yuncheng Qu, Lizhen Li, Zhuang Xue, Hao Salim, Flora D. Haffari, Gholamreza Department of Data Science & AI Monash University Australia School of Computer Science Engineering University of New South Wales Australia School of Computing Technologies Royal Melbourne Institute of Technology Australia

Alignment tuning is crucial for ensuring large language models (LLMs) behave ethically and helpfully. Current alignment approaches require high-quality annotations and significant training resources. This paper proposes a low-cost, tuning-free method using in-context learning (ICL) to enhance LLM alignment. Through an analysis of high-quality ICL demos, we identified style as a key factor influencing LLM alignment capabilities and explicitly restyled ICL exemplars based on this stylistic framework. Additionally, we combined the restyled demos to achieve a balance between the two conflicting aspects of LLM alignment—factuality and safety. We packaged the restyled examples as prompts to trigger few-shot learning, improving LLM alignment. Compared to the best baseline approach, with an average score of 5.00 as the maximum, our method achieves a maximum 0.10 increase on the Alpaca-eval task (from 4.50 → 4.60), a 0.22 enhancement on the just-eval-instruct benchmark (from 4.34 → 4.56), and a maximum improvement of 0.32 (from 3.53 → 3.85) on the MT-Bench dataset. We release the code and data at https://***/AnonymousCode-Computerscience/RIDE. © 2025, CC BY.

关键词： Contrastive Learning