Dr. Abdel Labbi

Dr. Abdel Labbi

IBM Research, Zurich, Switzerland

Abdel Labbi is an IBM Distinguished Engineer and Distinguished Research Staff Member (RSM), currently leading a global team of Engineers and Researchers building Enterprise scale AI Platforms. Since he joined IBM in 2000, he has led several teams and projects delivering Services & Software innovations that support both IBM internal business units and external clients in various industries. These innovations span the whole value chain of business, from Supply Chain Optimization to Sales & Marketing Effectiveness, Operations, Financial Planning, and Operational Risk Management. Prior to joining IBM, Dr. Labbi was Assistant Professor at the University of Geneva (Switzerland). He holds a PhD in applied Mathematics from the University of Grenoble (France), and an MBA from Henley Business School (UK).

Talk: Scalable AI Platforms for IoT Applications: Examples from Select Industries

Building scalable AI platforms is a significant endeavor that entails several challenges across the whole pipeline ranging from reliable data acquisition and transformation to AI models design, training and deployment in specific workflows/applications. A key challenge in such platforms is the trade-off between in-situ vs. remote data processing and AI models training. Recent developments both in Distributed Systems engineering and distributed AI models training make it possible to build reliable, yet high performance, IoT applications. We will show a few real examples in industries such as Transportation, Manufacturing, and Construction.

Prof. Keiichi Yasumoto

Prof. Keiichi Yasumoto

Graduate School of Science and Technology, NAIST, Japan

Keiichi Yasumoto is a Professor with the Graduate School of Science and Technology, Nara Institute of Science and Technology (NAIST). He has published a number of papers in mobile and ubiquitous computing areas, especially on participatory sensing systems, activity recognition in smart home, wireless sensor network and distributed edge computing platform. He served as a general co-chair of IEEE PerCom 2019 and participated in organization of many top-tier conferences including MobiQuitous, ICDCS, Pervasive, FORTE and MDM.

Talk: Toward Smarter Smarthome: Vision and Challenges

So far, many smart home technologies have been developed. Home appliances are now connected via a home network and dwellers can easily operate those appliances with a smart speaker, smartphone, and so on at home or even when out of the home. Technologies for recognizing activity of daily living (ADL) at home have also been developed. With the use of these technologies, many services such as automatic life logging and elderly care can be realized.

The ideal smart home would be a context-aware home, which recognizes contexts of both environments and dwellers and adapts environments to maximize the dweller’s QoL (Quality of Life). However, current smart home technologies are not yet matured enough to realize a true context-aware home. For instance, a context-aware home should be able to capture the dwellers’ intention as a context, e.g., the bed room should be set at a comfortable temperature just before he/she goes to bed, the oven should be pre-heated at the usual time he/she bakes pizza (cook food), and so on. To provide services to improve QoL of the dweller, technology to measure the dweller's QoL is needed. Once we could get dwellers’ detailed context including intention, QoL and so on, we can realize more sophisticated smart home services.

Meanwhile, the context-aware home should also be trustworthy for the dweller. For instance, it should not upload the sensor data from the home including high privacy information to external (untrusted) cloud servers unless granted by the dweller. This talk envisions a smarter smart home with several levels of smartness for context-awareness. In the end, we present several challenges including ADL forecasting, QoL measurement, and privacy protection technologies.