Class-Specific feature selection for enhanced cattle behavior recognition using wearable-based system

Yuxiao Yue

doi:10.1016/j.atech.2025.101548

2025-10-21

DOI：10.1016/j.atech.2025.101548

原文链接

查看原文

摘要

En 中文

对个体牛的行为进行监测和分析对于评估其健康状况和表现至关重要，从而能够提升现代畜牧生产管理。当前的行为识别系统通常依赖于可穿戴传感器和标准流程，该流程包括先进行特征提取，再进行特征选择以构建分类模型。然而，传统特征选择方法的一个关键局限性在于它们对所有行为类别使用统一的特征子集。这种方法无法捕捉类别特定的特征判别模式，并且随着行为类别的增加，可能导致分类性能下降。为了弥补这一差距，本文提出了一种类别特定特征选择（CSFS）框架，通过将原始C类行为识别任务分解为C个独立的二元分类子任务来重构多类行为识别。这使得能够根据每个行为类别的独特特征，为每个类别选择最优的特征子集。最终预测结果通过集成机制生成。实验比较和鲁棒性评估表明，所提出的框架相较于传统特征选择方法，平均准确率提升了21%（p<0.05），这一具有统计学显著性的增强显著提高了精准畜牧业应用中的识别可靠性。该框架兼容任意特征选择算法和机器学习分类器，并在噪声条件和不同传感器组合下表现出卓越的跨场景鲁棒性。通过提供更深层次的行为洞察并实现更准确的多行为并发识别，所提出的框架支持数据驱动的精准放牧系统的发展，并为可持续的畜牧管理实践做出贡献。

Keyword:

Cattle

Multi-behavior recognition

Class-specific feature selection

Machine learning

Wearable sensors

AI总结

对已上传原文的论文进行重点信息的提取，主要内容包括：简要概述、研究摘要、背景介绍、关键亮点、图文解析、展望与总结。

期刊

Smart Agricultural Technology

IF:

5.7

论文数:

2.5K

被引数:

2.5K

机构

Beijing Academy of Agriculture and Forestry Sciences

学者数:

1.1K

论文数: 348

被引数: 4

Jilin Agricultural University

学者数:

9.5K

论文数: 4.3K

被引数: 6.7K

引用论文

Applications of Savitzky-Golay Filter for Seismic Random Noise Reduction

Acta Geophysica

IF0

2016-02-01

OAAI

Yanping Liu; Bo Dang; Yue Li; Hongbo Lin; Haitao Ma

Classification of cow behavior patterns using inertial measurement units and a fully convolutional network model

JOURNAL OF DAIRY SCIENCE

IF4.4

2023-02-01

OAAI

Liu, Mei; Wu, Yiqi; Li, Guangyang; Liu, Meiqi; Hu, Rui; Zou, Huawei; Wang, Zhisheng; Peng, Yingqi

Adaptive Savitzky-Golay Filtering in Non-Gaussian Noise 非高斯噪声下的自适应savitzky-golay滤波

IEEE Transactions on Signal Processing

IF0

2021-01-01

PREAI

Arlene John; Jishnu Sadasivan; Chandra Sekhar Seelamantula

Real-Time Behavioral Recognition in Dairy Cows Based on Geomagnetism and Acceleration Information

IEEE ACCESS

IF3.6

2021-01-01

OAAI

Tian, Fuyang; Wang, Jun; Xiong, Benhai; Jiang, Linshu; Song, Zhanhua; Li, Fade

El Moutaouakil K, Falih N, Deep learning-based classification of cattle behavior using accelerometer sensors, IAES International Journal of Artificial Intelligence (IJ-AI) 13 (1) (2024) 524, 10.11591/ijai.v13.i1.pp524-532.El Moutaouakil K, Falih N, 基于深度学习的牛行为分类方法研究：基于加速度传感器，IAES 国际人工智能杂志（IJ-AI）13（1）（2024）524, 10.11591/ijai.v13.i1.pp524-532.

IF0

2024-03-01

PREAI

Khalid El Moutaouakil; Noureddine Falih

Evaluation of the Efficiency of Machine Learning Algorithms for Identification of Cattle Behavior Using Accelerometer and Gyroscope Data

AGRIENGINEERING

IF3

2024-07-16

OAAI

Mladenova, Tsvetelina; Valova, Irena; Evstatiev, Boris; Valov, Nikolay; Varlyakov, Ivan; Markov, Tsvetan; Stoycheva, Svetoslava; Mondeshka, Lora; Markov, Nikolay

A Deep Learning Approach for Detecting and Classifying Cat Activity to Monitor and Improve Cat's Well-Being Using Accelerometer, Gyroscope, and Magnetometer

IEEE SENSORS JOURNAL

IF4.5

2024-01-15

PREAI

Hussain, Ali; Ali, Sikandar; Joo, Moon-Il; Kim, Hee-Cheol

V V, Sb S, Laxmy M, Monisha S, Sanjey A, P A, A Class Specific Feature Selection Method for Improving the Performance of Text Classification, Scalable Computing Practice and Experience 25 (2024) 1018-28, 10.12694/scpe.v25i2.2502.V V, Sb S, Laxmy M, Monisha S, Sanjey A, P A, 针对提升文本分类性能的类别特定特征选择方法，Scalable Computing Practice and Experience 25 (2024) 1018-28, 10.12694/scpe.v25i2.2502.

IF0

2024-02-24

PREAI

Venkatesh V; Sharan S B; Mahalaxmy S; Monisha S; Ashick Sanjey D S; Ashokkumar P

Feature redundancy term variation for mutual information-based feature selection