Industrial AI is concerned with the application of Artificial Intelligence (AI), Machine Learning (ML) and related technologies towards addressing real-world use cases in industrial and societal domains. These use cases can be broadly categorized into the horizontal areas of: maintenance & repair, operations & supply chain, quality, safety, design, and end-to-end optimization - with applications in a variety of verticals. In the last few years, we have witnessed a growing interest in applying Deep Learning (DL) techniques to Industrial AI problems, ranging from using sequence models such as Long Short-Term Memory (LSTM) for predicting failures in equipment, to using Deep Reinforcement Learning (Deep RL) for scheduling and dispatching. Applying DL techniques to industrial applications imposes a set of unique challenges, which include, but are not limited to, (1) limited data, highly skewed class distribution and occurrence of rare classes such as failures, (2) multi-modal data (sensors, events, images, text, etc.) indexed over space and time (3) the need for explainable decisions, (4) a need to attain consistency between different but “related” models and between multiple generations of the same model, and (5) decision making to optimize business outcomes where the cost of a mistake could be very high. This tutorial presents an overview of these challenges, along with new methods and best practices to address them. Examples of these methods include using sequence DL models and Functional Neural Networks (FNNs) for modeling sensor and spatiotemporal measurements; using multi-task learning, graph models and ensemble learning for improving consistency of DL models; using deep RL for health indicator learning and dynamic dispatching; cost-based decision making for prognostics; and using GANs for generating senor data for prognostics. Finally, we will present some open problems in Industrial AI and how the research community can shape the future of the next industrial and societal revolution.

Tutors

Chetan Gupta, Chief Data Scientist and Head of Industrial AI Lab, Hitachi America, Ltd. R&D

Ahmed Farahat, Principal Research Scientist, Industrial AI Lab, Hitachi America, Ltd. R&D

Slides

Video

Outline

Part I: Introduction to Industrial AL

1. Introduction and Motivation.

• Industrial AI: Introduction and motivation

• Data sources, business problems, challenges, and restrictions

2. Problems and Formulations.

• Taxonomy of Industrial AI problems

• Problem formulations and modeling challenges

Part II: Traditional Formulations: From Machine to Deep Learning

3. Deep Learning for Prognostics.

• Introduction to Prognostics

• Prognostics using sequence deep learning models [5, 8, 22]

• Functional neural networks for modeling sensors and spatiotemporal measurements [10, 14, 15]

• Prognostics in industrial networks using graph CNN [4, 7]

• Generative adversarial networks for prognostics and sensor generation [18–20]

4. Deep Learning for Perception and Natural Language Understanding.

• Visual inspection for maintenance, manufacturing and quality enhancement [3, 13]

• Natural language understanding and knowledge base construction for industrial domains [2, 11]

Part III: Practical Perspectives and New Formulations

5. Attaining Consistency between Related Models.

• Multi-task learning for achieving consistency between related tasks [1]

• Ensemble learning for improving consistency of deep learning models [6, 9]

6. Deep Reinforcement Learning for Industrial Applications.

• Deep RL for dynamic dispatching [21]

• Health indicator learning using deep RL [16]

Part IV: Future of Industrial AI and Conclusions

7. Open Problems, Future of Industrial AI and Conclusions

References

[1] Karan Aggarwal, Onur Atan, Ahmed K Farahat, Chi Zhang, Kosta Ristovski, and Chetan Gupta. 2018. Two birds with one network: Unifying failure event prediction and time-to-failure modeling. In 2018 IEEE International Conference on Big Data (Big Data). IEEE, 1308–1317.

[2] Aravind Chandramouli, Gopi Subramanian, Debasis Bal, SI Ao, C Douglas, WS Grundfest, and J Burgstone. 2013. Unsupervised extraction of part names from service logs. In Proceedings of the World Congress on Engineering and Computer Science, Vol. 2.

[3] Sreerupa Das, Christopher D Hollander, and Suraiya Suliman. 2019. Automating Visual Inspection with Convolutional Neural Networks. In Annual Conference of the PHM Society, Vol. 11.

[4] Michaël Defferrard, Xavier Bresson, and Pierre Vandergheynst. 2016. Convolutional neural networks on graphs with fast localized spectral filtering. In Advances in neural information processing systems. 3844–3852.

[5] Wei Huang, Hamed Khorasgani, Chetan Gupta, Ahmed Farahat, and Shuai Zheng. 2018. Remaining useful life estimation for systems with abrupt failures. In Annual Conference of the PHM Society, Vol. 10.

[6] Riashat Islam, Peter Henderson, Maziar Gomrokchi, and Doina Precup. 2017. Reproducibility of benchmarked deep reinforcement learning tasks for continuous control. arXiv preprint arXiv:1708.04133 (2017).

[7] Hamed Khorasgani, Arman Hasanzadeh, Ahmed Farahat, and Chetan Gupta. 2019. Fault detection and isolation in industrial networks using graph convolutional neural networks. In 2019 IEEE International Conference on Prognostics and Health Management (ICPHM). IEEE, 1–7.

[8] Pankaj Malhotra, Vishnu TV, Anusha Ramakrishnan, Gaurangi Anand, Lovekesh Vig, Puneet Agarwal, and Gautam Shroff. 2016. Multi-sensor prognostics using an unsupervised health index based on LSTM encoder-decoder. In 1st ACM SIGKDD Workshop on Machine Learning for Prognostics and Health Management.

[9] Prasad Patil and Giovanni Parmigiani. 2018. Training replicable predictors in multiple studies. Proceedings of the National Academy of Sciences 115, 11 (2018).

[10] Fabrice Rossi, Brieuc Conan-Guez, and François Fleuret. 2002. Functional data analysis with multi layer perceptrons. In Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN’02 (Cat. No. 02CH37290), Vol. 3. IEEE, 2843–2848.

[11] Walid Shalaby, Adriano Arantes, Teresa GonzalezDiaz, and Chetan Gupta. 2020. Building chatbots from large scale domain-specific knowledge bases: challenges and opportunities. In 2020 IEEE international conference on prognostics and health management (ICPHM). IEEE.

[12] Stephan Spiegel, Fabian Mueller, Dorothea Weismann, and John Bird. 2018. Cost-sensitive learning for predictive maintenance. arXiv preprint arXiv:1809.10979 (2018).

[13] Domen Tabernik, Samo Šela, Jure Skvarč, and Danijel Skočaj. 2020. Segmentation-based deep-learning approach for surface-defect detection. Journal of Intelligent Manufacturing 31, 3 (2020), 759–776.

[14] Qiyao Wang, Shuai Zheng, Ahmed Farahat, Susumu Serita, and Chetan Gupta. 2019. Remaining useful life estimation using functional data analysis. In 2019 IEEE International Conference on Prognostics and Health Management (ICPHM). IEEE, 1–8.

[15] Qiyao Wang, Shuai Zheng, Ahmed Farahat, Susumu Serita, Takashi Saeki, and Chetan Gupta. 2019. Multilayer perceptron for sparse functional data. In 2019 International Joint Conference on Neural Networks (IJCNN). IEEE, 1–10.

[16] Chi Zhang, Chetan Gupta, Ahmed Farahat, Kosta Ristovski, and Dipanjan Ghosh. 2018. Equipment health indicator learning using deep reinforcement learning. In Joint European Conference on Machine Learning and Knowledge Discovery in Databases. Springer, 488–504.

[17] Chi Zhang, Chetan Gupta, Seiji Joichi, Ahmed Farahat, and Huijuan Shao. 2019. Risk-Based Dynamic Pricing via Failure Prediction. In 2019 18th IEEE International Conference On Machine Learning And Applications (ICMLA). IEEE, 140–147.

[18] Shuai Zheng, Ahmed Farahat, and Chetan Gupta. 2019. Generative adversarial networks for failure prediction. In Joint European Conference on Machine Learning and Knowledge Discovery in Databases. Springer, 621–637.

[19] Shuai Zheng and Chetan Gupta. 2020. Discriminant Generative Adversarial Networks with its Application to Equipment Health Classification. In ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 3067–3071.

[20] Shuai Zheng and Chetan Gupta. 2020. Trace Norm Generative Adversarial Networks for Sensor Generation and Feature Extraction. In ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 3187–3191.

[21] Shuai Zheng, Chetan Gupta, and Susumu Serita. 2019. Manufacturing Dispatching using Reinforcement and Transfer Learning. In Joint European Conference on Machine Learning and Knowledge Discovery in Databases. Springer, 655–671.

[22] Shuai Zheng, Kosta Ristovski, Ahmed Farahat, and Chetan Gupta. 2017. Long short-term memory network for remaining useful life estimation. In 2017 IEEE international conference on prognostics and health management (ICPHM). IEEE, 88–95.