CORPORATE TALK SERIES
(Chief Scientist and Co-Founder, Coherent Knowledge)
Bio: Benjamin Grosof, PhD, is Co-Founder and Chief Scientist at Coherent Knowledge, an AI startup that provides highly explainable decision support via query answering. He has pioneered technology and industry standards for knowledge graphs and expressively flexible semantic rules, their acquisition from natural language, how to combine them with machine learning, and a wide variety of applications including in defense & security, finance, legal & policy, e- commerce & supply chain, health care & life science, and helpdesk. In particular, he has led invention of declarative logic programs that extend databases scalably with powerful implications and meta-logic features such as probabilities, higher-order syntax, bounded rationality (restraint), exceptions/argumentation (defeasibility), and user-friendly explanations. Previously, he was a technical/research executive in AI at: the Allen Institute for AI’s predecessor; Accenture; and Kyndi, a venture-backed AI startup. Earlier, he was a MIT Sloan professor and DARPA PI, and an IBM Research scientist. His background includes a part-time expert consulting practice, Stanford PhD in computer science (specialty AI), Harvard BA in applied mathematics, 60+ refereed publications, 10,000+ citations, 5 patents, 2 W3C standards, and 5 major industry software products.
Title of the Talk: AI’s Secret Weapon for Ubiquitous Computing: Extended Logic Programs
Abstract: Hybrid AI, including neuro-symbolic, which combines machine learning (ML) with logical knowledge representation & reasoning (KRR), is at the forefront of innovation in AI. This talk focuses on how and why hybrid AI will be crucial to the future development of ubiquitous computing.
We focus specifically on extended declarative logic programs (ELP) as a key, yet under-appreciated, KRR technique for semantics within hybrid AI that can aid a critical requirement for ubiquitous computing: interoperability. By supporting a high level of abstraction, ELP can increase portability across the panoply of data, device, network, and application task settings. Furthermore, ELP can improve the familiarity, ease, and depth of user interaction by non-programmers -- including via natural language together with high explainability -- a second critical objective in many ubiquitous computing applications. ELP can thereby reduce cost and increase agility, in both development and operation. One particularly promising direction is hybrid AI that combines ELP with neural net based natural language processing, for conversational assistants.
ELP has already been a “hidden” common basis for design of existing web standards for semantics in knowledge graphs and ontologies, which have recently become widely adopted in industry, increasingly for ubiquitous computing applications. But as part of hybrid AI, ELP has the potential to reach much further. We illustrate ELP’s broad capabilities with examples using a state-of-the-art ELP system, ErgoAI from Coherent Knowledge, that can represent knowledge quite flexibly, reason powerfully at scale, and provide full explanations that are highly understandable even by non-programmers, so as to answer queries and support decisions, while orchestrating many other network-based services and components, e.g., data sources, streams, and ML programs. ELP can help represent and reason about the semantics and implications in both structured knowledge (e.g., databases or spreadsheets) and unstructured knowledge (e.g., text). ELP shines when a high degree of assurance or trust is required, e.g., in specification and execution of operational policies or legal terms & conditions (such as in business processes, contracts, regulations, security & privacy), which are an important part of many ubiquitous computing applications. There are exciting opportunities to leverage ELP in hybrid AI in diverse ubiquitous computing domains including e-commerce, supply chain, customer care, health, smart home, automotive, defense, and financial.
(Principal Scientist, Bristol Myers Squibb)
Bio: Celine Han is a computational biologist and principal scientist in Translational Bioinformatics at Bristol Myers Squibb (BMS). Her focus of research is to advance oncology clinical trial programs and to influence development strategies by applying cutting-edge bioinformatic analyses on clinical trials and real-world data. Her work involves driving complex collaborations and communications with various partners for biomarker discovery and patient stratification. Prior to joining BMS, she finished her postdoctoral training at Harvard Medical School, Dana- Farber Cancer Institute and Broad Institute of MIT and Harvard. She studied the resistance to hormone therapy for patient stratification in metastatic prostate cancer. She completed her Ph.D. training in Bioinformatics and Genomics at Pennsylvania State University studying the epigenetics and transcriptional regulation in erythropoiesis. Outside of research, she is passionate about science communication, mentoring, teaching, and being a lifelong learner.
|Full Paper Submission:||7th November 2021|
|Acceptance Notification:||17th November 2021|
|Final Paper Submission:||24th November 2021|
|Early Bird Registration||21st November 2021|
|Presentation Submission:||23rd November 2021|
|Conference:||1 - 4 December 2021|
• Conference Proceedings will be submitted for publication at IEEE Xplore® digital library.
• Best Paper Award will be given for each track.
• Conference Record No 53757