Yutian Ren

Machine learning (ML) has been demonstrated to improve productivity in many manufacturing applications. To host these ML applications, several software and Industrial Internet of Things (IIoT) systems have been proposed for manufacturing applications to deploy ML applications and provide real-time intelligence. Recently, an interactive causality-enabled self-labeling method has been proposed to advance adaptive ML applications in cyber–physical systems, especially manufacturing, by automatically adapting and personalizing ML models after deployment to counter data distribution shifts. The unique features of the self-labeling method require a novel software system to support dynamism at various levels. This paper proposes the AdaptIoT system, comprising an end-to-end data streaming pipeline, ML service integration, and an automated self-labeling service. The self-labeling service consists of causal knowledge bases and automated full-cycle self-labeling workflows to adapt multiple ML models simultaneously. AdaptIoT employs a containerized microservice architecture to deliver a scalable and portable solution for small and medium-sized manufacturers. A field demonstration of a self-labeling adaptive ML application is conducted with a makerspace and shows reliable performance with comparable accuracy at 98.3%.

Machine Learning (ML) has been demonstrated to improve productivity in many manufacturing applications. To host these ML applications, several software and Industrial Internet of Things (IIoT) systems have been proposed for manufacturing applications to deploy ML applications and provide real-time intelligence. Recently, an interactive causality enabled self-labeling method has been proposed to advance adaptive ML applications in cyber-physical systems, especially manufacturing, by automatically adapting and personalizing ML models after deployment to counter data distribution shifts. The unique features of the self-labeling method require a novel software system to support dynamism at various levels. This paper proposes the AdaptIoT system, comprised of an end-to-end data streaming pipeline, ML service integration, and an automated self-labeling service. The self-labeling service consists of causal knowledge bases and automated full-cycle self-labeling workflows to adapt multiple ML models simultaneously. AdaptIoT employs a containerized microservice architecture to deliver a scalable and portable solution for small and medium-sized manufacturers. A field demonstration of a self-labeling adaptive ML application is conducted with a makerspace and shows reliable performance.

Adaptive machine learning (ML) aims to allow ML models to adapt to ever-changing environments with potential concept drift after model deployment. Traditionally, adaptive ML requires a new dataset to be manually labeled to tailor deployed models to altered data distributions. Recently, an interactive causality based self-labeling method was proposed to autonomously associate causally related data streams for domain adaptation, showing promising results compared to traditional feature similarity-based semi-supervised learning. Several unanswered research questions remain, including self-labeling's compatibility with multivariate causality and the quantitative analysis of the auxiliary models used in the self-labeling. The auxiliary models, the interaction time model (ITM) and the effect state detector (ESD), are vital to the success of self-labeling. This paper further develops the self-labeling framework and its theoretical foundations to address these research questions. A framework for the application of self-labeling to multivariate causal graphs is proposed using four basic causal relationships, and the impact of non-ideal ITM and ESD performance is analyzed. A simulated experiment is conducted based on a multivariate causal graph, validating the proposed theory.

Learning from unlabeled data, or self-learning, can substantially reduce the complexity of machine learning (ML) utilization in real-time deployment. While the development of un/ semi-supervised algorithms shows promising results in learning with reduced labels, the fundamental assumption of data smoothness restricts its scope of application, especially with non-stationary data distributions in different domains. Leveraging trans-domain invariant causal relationships, causality has recently been employed to foster robust ML. In this paper, we have developed a generic method for self-labeling data that relies on known causality among interactive objects and learned temporal relations among causal events for identifying and associating labels and input data. The causal time-interval between asynchronous cause and effect events is studied to achieve self-labeling. We utilize dynamical system theory in a 1- $d$ setting to demonstrate that our proposed method outperforms traditional feature similarity based semi-supervised learning. A computer simulation experiment was conducted to generate high-dimensional data, and the comprehensive results reveal the potential of learning adaptation in dynamic environments to improve ML robustness against shifts in data distribution.

An adaptive machine learning (ML) based smart manufacturing interactive cyber physical human system (ICPHS) is conceptualized, designed, and implemented. One of its significant properties is an ML model that during deployment self-evolves with the streaming data in a self-labeling manner. This automated adaptive ML system is realized by leveraging the underlying causality during human machine interactions, and initialized by a priori domain knowledge and preprocessed public datasets. The system defines a causal and temporal mapping of worker and machine states where one side can label the other automatically. A case study in machines with different automation levels is conducted in a multiuser facility by using finite state machine representations for machine energy states, worker activity states, and interaction transition functions. The fully automated adaptive ML system improves its accuracy for human machine interaction detection by up to 12.5% and shows potential to recognize more fine-grained actions.