China
Africa
Explore chapters and articles related to this topic
Decision Support Systems for Anomaly Detection with the Applications in Smart Manufacturing: A Survey and Perspective
Published in Kim Phuc Tran, Machine Learning and Probabilistic Graphical Models for Decision Support Systems, 2023
Quoc-Thông Nguyen, Tung Nhi Tran, Cédric Heuchenne, Kim Phuc Tran
In production monitoring, to reduce human intervention and maintain a high level of accuracy in the process, the advanced techniques need to eliminate most of the assumptions of traditional SPC and focus on complex data types (multivariate time series data, image data, and Big Data with complex structures). The FL approaches will be based on advanced DL algorithms such as long short-term memory (LSTM) 80, convolution neural network81, Hierarchical Temporal Memory (HTM)4 combined with anomaly detection methods 82;5, kernel density estimation (KDE) 83, e.g., AEs, for real-time multivariate time series data and image data monitoring.
A survey of intrusion detection from the perspective of intrusion datasets and machine learning techniques
Published in International Journal of Computers and Applications, 2022
In humans, the neocortex performs complex tasks such as object identification and manipulation by touch, visual pattern recognition, spoken language comprehension, etc. HTM technology implements human-level cognitive tasks performed by the neocortex [38]. HTM is a time-based continuous learning algorithm similar to the neural network. HTM store and recall spatial and temporal patterns. It scores anomalies between 0 and 1. Mean and variance of the latest anomaly scores distribution is maintained by the system to correlate the current anomaly score with respective normal distribution at every time step. The final anomaly is predicted by thresholding the anomaly likelihood. The Numenta Platform for Intelligent Computing (NuPIC) provides a platform to implement the HTM learning algorithms. HTM attains significant performance improvements on NAB benchmarks. But it does not perform so well on the Yahoo Webscope S5 dataset.