The annual SIGKDD conferences have a strong reputation for delivering unparalleled peer learning, networking and idea sharing opportunities within data science, data mining, knowledge discovery, large-scale data analytics and big data. The event’s main focus is to connect the world’s best data scientists with one another in order to discuss, address and advance the application of data science to benefit all aspects of society. KDD 2020 is looking to be an amazing year.
A Look at State-Space Multi-Taper Time-Frequency Analysis
Time series arising for studies of physical, biological, economic and sociological systems are an important data class. The growing interest in this type of data has come about because of significant recent advances in sensor, recording and digitization technologies. An increasingly common objective is accurate, real-time analyses of these non-stationary series. At present, a mainstay of these analyses is the use of time-varying spectral methods that rely on the multi-taper (MT) technique, applied across overlapping or non-overlapping data windows. This paradigm provides a useful, widely applicable approach to time-series analysis. However, MT spectral analysis is only optimal for spectral estimation on a local interval and it does not offer a framework for formal statistical inference that can be applied to an entire time series. In this lecture, we will discuss our recent work on the development of a state-space multi-taper (SS-MT) framework for the analysis of non-stationary time series. The non-stationary time-series is represented as a union of locally stationary intervals. We present a frequency-domain state-space model based on the time series spectral representation to link the local intervals. The model parameters can be efficiently implemented using an EM algorithm whereas the spectral updates are efficiently computed using a parallel set of 1D complex Kalman filter algorithms. We illustrate the improved denoising and enhanced resolution of the SS-MT approach relative to the MT methods in a simulation study and in the analysis of actual EEG time-series recordings in anesthetized patients. We further demonstrate SS-MT’s empirical Bayes’ inference framework using a second set of EEG recordings from anesthetized patients. We believe the SS-MT approach offers several possibilities for improving analyses of non-stationary time-series.
Bio: Emery N. Brown is the Edward Hood Taplin Professor of Medical Engineering and Computational Neuroscience at MIT; the Warren M. Zapol Professor of Anaesthesia at Harvard Medical School; and a practicing anesthesiologist at Massachusetts General Hospital. His experimental research has helped define the neuroscience mechanisms of how anesthetics work. His statistics research has developed a broad range of statistical and signal processing methods to improve neuroscience data analysis. He is a fellow of the IEEE, the American Statistical Association, the Institute for Mathematical Statistics, and the American Association for the Advancement of Science. Professor Brown is the recipient of the Dickson Prize in Science and an honorary Doctorate of Science degree from U.S.C. He is a member of the American Academy of Arts Sciences, the National Academy of Inventors, National Academy of Medicine, National Academy of Sciences and National Academy of Engineering.
AI for Intelligent Financial Services: Examples and Discussion
There are many opportunities to pursue AI and ML in the financial domain. In this talk, I will overview several research directions we are pursuing in engagement with the lines of business, ranging from data and knowledge, learning from experience, reasoning and planning, multi agent systems, and secure and private AI. I will offer concrete examples of projects, and conclude with the many challenges and opportunities that AI can offer in the financial domain.
Bio: Manuela M. Veloso is the Head of J.P. Morgan AI Research, which pursues fundamental research in areas of core relevance to financial services, including data mining and cryptography, machine learning, explainability, and human-AI interaction. J.P. Morgan AI Research partners with applied data analytics teams across the firm as well as with leading academic institutions globally. Professor Veloso is on leave from Carnegie Mellon University, where she has been a professor for over 30 years. Professor Veloso is the Past President of AAAI, (the Association for the Advancement of Artificial Intelligence), and the co-founder, Trustee, and Past President of RoboCup. Professor Veloso has been recognized with a multiple honors, including being a Fellow of the ACM, IEEE, AAAS, and AAAI. She is the recipient of several best paper awards, the Einstein Chair of the Chinese Academy of Science, the ACM/SIGART Autonomous Agents Research Award, an NSF Career Award, and the Allen Newell Medal for Excellence in Research. See www.cs.cmu.edu/~mmv/Veloso.html for her publications.
Generating Explanations that Matter through Meta-Provenance
Provenance standards have now been used for many years to generate useful explanations of the data analytic process used to generate a new finding. These explanations convey the details of analytic steps and the original data used in an analysis. In this talk, I will discuss the need for explanations that provide the context and rationale for how the data analysis process was designed. I will show how to capture meta-provenance that can be used to generate those explanations. I will present our work to date on capturing method abstractions and generalizations, templates that ensure that analyses follow proven approaches, questions that drive many analytic choices, and detailed metadata about models and their differences. I will illustrate with examples from several domains the kinds of explanations that can be generated from meta-provenance, and discuss important areas of future work.
Bio: Dr. Yolanda Gil is Director for Major Strategic AI and Data Science Initiatives at the Information Sciences Institute of the University of Southern California, and Research Professor in Computer Science and in Spatial Sciences. She is also Director for Data Science Programs, and Director of the USC Center for Knowledge-Powered Interdisciplinary Data Science. She received her M.S. and Ph. D. degrees in Computer Science from Carnegie Mellon University, with a focus on artificial intelligence. Her research is on intelligent interfaces for knowledge capture and discovery, which she investigates in a variety of projects concerning scientific discovery, knowledge-based planning and problem solving, information analysis and assessment of trust, semantic annotation and metadata, and community-wide development of knowledge bases. Dr. Gil collaborates with scientists in many domains on semantic workflows and metadata capture, social knowledge collection, computer-mediated collaboration, and automated discovery. She is a Fellow of the Association for Computing Machinery (ACM), and Past Chair of its Special Interest Group in Artificial Intelligence. She is also Fellow of the Association for the Advancement of Artificial Intelligence (AAAI), and was elected as its 24th President in 2016.
Computational Epidemiology at the time of COVID-19
The data science revolution is finally enabling the development of large-scale data-driven models that provide scenarios, forecasts and risk analysis for infectious disease threats. These models also provide rationales and quantitative analysis to support policy making decisions and intervention plans. At the same time, the non-incremental advance of the field presents a broad range of challenges: algorithmic (multiscale constitutive equations, scalability, parallelization), real time integration of novel digital data streams (social networks, participatory platform, human mobility etc.). I will review and discuss recent results and challenges in the area and focus on ongoing work aimed at responding to the COVID-19 pandemic.
Bio: Alessandro Vespignani research activity is focused on the study of “techno-social” systems, where infrastructures composed of different technological layers are interoperating within the social component that drives their use and development. In this context we aim at understanding how the very same elements assembled in large number can give rise – according to the various forces and elements at play – to different macroscopic and dynamical behaviors, opening the path to quantitative computational approaches and forecasting power.
How can we assist you?
We'll be updating the website as information becomes available. If you have a question that requires immediate attention, please feel free to contact us. Thank you!
Please enter the word you see in the image below: