IEEE Green Energy and Smart Systems Conference
Long Beach, CA

Lunch Speaker

Topic: Toward greener and Cleaner Campus

Abstract and Bio:

Dr. Jane Close Conoley serves as the seventh President of Long Beach State University.  Prior to assuming this role she was the interim Chancellor of the University of California Riverside.  She has also held leadership positions at the University of California Santa Barbara, Texas A&M University, and at the University of Nebraska Lincoln.  Dr. Conoley is the author, co-author or editor of over one hundred thirty books, articles, and book chapters.   Her latest book “Positive Psychology and Family Therapy” was co-authored with her husband, Dr. Collie W. Conoley.

She serves on numerous journal editorial and community service boards.  She has received research, teaching and service honors during her career.  Both the American Psychological Association and the Association of Psychological Science honored Dr. Conoley with fellow status.  Her research and development efforts in school safety, teacher quality, and student achievement have been supported by over 50 million dollars in external federal, state, and private funds.

Topic: Analytics for Industrial Internet Applications

Abstract: The Industrial Internet is the third disruptive wave, after the Industrial and the Internetrevolutions. It is transforming our industries, just like the Internet revolution transformed our commerce.  In this new context, we face a combination of hyper-connected intelligent machines, interacting with other machines and people, and generating large amounts data that need to be analyzed by descriptive, predictive, and prescriptive models.  As a result, we see the resurgence of analytics as a key differentiator for creating new services, the emergence of cloud computing as an enabling technology for service delivery, and the growth of crowdsourcing as a new phenomenon in which people play critical roles in creating information and shaping decisions in a variety of problems. We explore the intersection of these three concepts from the perspective of a machine-learning researcher and show how his job and roles have evolved over time.

In the past, analytic model creation was an artisanal process, as models were handcrafted by experienced, knowledgeable model-builders. More recently, the use of meta-heuristics, such as evolutionary algorithms, has provided us with limited levels of automation in model building and maintenance.  In the short future, we expect data-driven analytic models to become a commodity. We envision having access to a large number of data-driven models, obtained by a combination of crowdsourcing, cloud-based evolutionary algorithms, outsourcing, in-house development, and legacy models. In this context, the critical issue will be model ensemble selection and fusion, rather than model generation.

First, we will review the application of data-driven analytic models to assets diagnostics and prognostics, such as aircraft engines, medical imaging devices, and locomotives.  We will cover a case study on prediction of remaining useful life for each unit in a fleet of locomotives using fuzzy models.

Then we will explore the evolution of analytic models with the advent of cloud computing, and propose the use of customized model ensembles on demand, inspired by Lazy Learning. This approach is agnostic with respect to the origin of the models, making it scalable and suitable for a variety of applications.  We successfully tested this approach in a regression problem for a power plant management application, using two different sources of models: bootstrapped neural networks, and GP-created symbolic regression models evolved in the cloud. We will also present results on the fusion of models for FlyQuest, a GE-sponsored Kaggle competition in which we crowd-sourced the generation of models predicting the estimated runway and gateway arrival (ERA, EGA) over a month of US flights.

Finally, we will explore research trends, challenges and opportunities for Machine Learning techniques in this emerging context of big data and cloud computing.



Dr. Piero P. Bonissone is an independent consultant specialized in the use of analytics for Industrial Internet applications. He provides consulting services in machine learning (ML) and analytic applications, ranging from project definition and risk abatement, project evaluation, transition from development to deployment, and model maintenance. During 2018, he was an Advanced Analytics advisor for Stanley Black Decker (SBD). In 2017, he defined and shaped new projects for GE Oil & Gas, prior to their integration with Baker Hughes Inc. (BHI). During the previous two years, he was an Advanced Analytics Advisor for Schlumberger (SLB), where he played a key role in SLB Digital Transformation, such as part forecasting, market intelligence, PHM projects related to equipment reliability, etc. He was also a consultant for DIGILE and Ford Motor Co.

A former Chief Scientist at GE Global Research (GE GR), where he retired in 2014 after 34 years of service, Dr. Bonissone has been a pioneer in the field of analytics, machine learning, fuzzy logic, AI, and soft computing applications.

He is a Fellow of the Institute of Electrical and Electronics Engineers (IEEE), the Association for the Advancement of Artificial Intelligence (AAAI), the International Fuzzy Systems Association (IFSA), and a Coolidge Fellow at GE Global Research. He received the 2012 Fuzzy Systems Pioneer Award from the IEEE Computational Intelligence Society (CIS). From 2010 to 2015, he chaired the Scientific Committee of the European Centre for Soft Computing. In 2008 he received the II Cajastur International Prize for Soft Computing from the European Centre of Soft Computing. In 2005 he received the Meritorious Service Award from the IEEE CIS. He has received two Dushman Awards from GE Global Research. He served as Editor-in-Chief of the International Journal of Approximate Reasoning for 13 years. He is in the editorial board of five technical journals and is Editor at Large of the IEEE Computational Intelligence Magazine. He co-edited six books and has 150+ publications in refereed journals, book chapters, and conference proceedings, with 10,000+ citations, an H-Index of 52 and an i10-index of 160 (by Google Scholar). He received 73 patents issued by the US Patent Office (and 10+ pending patents). From 1982 until 2005 he has been an Adjunct Professor at Rensselaer Polytechnic Institute, in Troy NY, where he supervised 5 PhD theses and 34 Master theses. He co-chaired 12 scientific conferences focused on Multi-Criteria Decision-Making, Fuzzy sets, Diagnostics, Prognostics, and Uncertainty Management in AI. He has been a member of the IEEE Fellow Committee in 2007-09; 2012-14, and 2016-19. In 2002, while serving as President of the IEEE Neural Networks Society (now CIS), he was a member of the IEEE TAB.. He was an ExCom/AdCom member of NNC/NNS/CIS society in 1993-2012 and 2016-18 and an IEEE CIS Distinguished Lecturer in 2004-14, and in 2017-19.


Topic: A Walk with Neural Networks: Interesting Behaviors in Training, Inference, Pruning, and Controlling

Abstract: In the recent decade, new advancement in AI has been largely enabled by the development of neural networks, in computer vision, natural language processing, reinforcement learning, robotics, and many other areas. While many neural networks deliver superior performance at their tasks, networks are fundamentally complex systems, and their training and operation is still poorly understood. In this talk, imagine we take a walk with the fantastic beasts — neural networks, and have conversations around a number of active topics dedicated to the understanding of their behaviors.

How is the training process like under microscopic inspection — how to measure each parameter’s contribution to loss change?

How to define and measure intrinsic network complexity, and what does that measure tell us?

How to uncover hidden flaws in a popular model, and is there a better alternative?

How to speed up both training and inference, from more natural input spaces of images?

How to deconstruct the mystical success in pruning networks to develop better architectures and initializations, and what are the implications?

And lastly, we will cover a lightweight, flexible control mechanism in language generation with large transformer-based models.




Dr. Rosanne Liu is a senior research scientist, and a founding member of Uber AI. She obtained her PhD in Computer Science at Northwestern University, where she used neural networks to help discover novel materials. She is currently working on the multiple fronts where machine learning and neural networks are mysterious. She has multiple publications at NeurIPS, ICLR, ICML, and other top machine learning venues, and her work had been featured in MIT Tech Review, WIRED, TechCrunch and Fortune. She was named 30 Rising stars in AI by ReWork in 2019, and 30 Influential Women Advancing AI in San Francisco in 2018.