Aim

This workshop aims to catalyze fundamental research collaborations between MIT, National Institute of Informatics, and Japanese researchers, at the intersection of computer science and artificial intelligence, including applications of broad civic importance. It will also feature partnerships between MIT, Keio University, and Chiba Institute of Technology, and pioneering open-source probabilistic computing systems engineering by Digital Garage in Japan.

   Program

June 13th, 2022

15:30 - 15:35 Opening
Joi Ito (Chiba Institute of Technology)
Kitsuregawa Masaru (National Institute of Informatics)
15:35 – 16:05 On an alternate route to scaling AI via probabilistic programming
Vikash Mansinghka (MIT Probabilistic Computing Project)

This talk will introduce fundamentals of probabilistic programming, including new probabilistic programming-based AI moonshots being pursued by MIT in collaboration with Google and Meta. Capabilities will be grounded in new SOTA results in indoor 3D scene perception, large-scale medical database linkage, automated modeling of clinical trials, and automated econometric forecasting.
It will also briefly review emerging economic opportunities and security risks arising from the surprising robustness, data efficiency, and compute efficiency of probabilistic programs, as compared to current neural networks on their own.

16:05 – 16:20 InferenceQL: making probabilistic programming usable by ~10 million SQL users
Ulrich Schaechtle (Digital Garage)
16:20 – 16:40 Fundamental limits on scaling and automation for probabilistic computing
Cameron Freer (Keio University & MIT Probabilistic Computing Project)

This talk will review fundamental concepts and theoretical results relevant for four topics: computability of probabilistic inference; the efficiency of random variate generation; prospects for decentralized, planetary-scale probabilistic computation; and prospects for using probabilistic programs to learn to generate synthetic data.

break
17:30 – 17:50 Average sensitivity analysis of optimization algorithms
Yuichi Yoshida (National Institute of Informatics)

As computing power increases, algorithms are increasingly used in decision making. However, many existing algorithms significantly change their outputs when inputs change or noise is added. As a result, the decisions are also subject to frequent changes, which can result in large costs and loss of trust. To cope with this situation, the speakers proposed a concept called "average sensitivity of algorithms" and have proposed algorithms with small average sensitivity for various problems such as graph-related problems, clustering, and decision tree learning. In this talk, I will introduce recent progress on average sensitivity of algorithms.

June 15th, 2022

15:00 - 15:20 Geometric learning with graph structure
Mahito Sugiyama (National Institute of Informatics)

This talk will introduce a close connection between dually flat manifolds, which is known as the canonical structure in information geometry, and incidence algebras in order theory, and present its applications to machine learning. This approach allows us to flexibly design log-linear models on a directed acyclic graph, which include a number of machine learning problems such as learning of Boltzmann machines, matrix/tensor low-rank approximation, and blind source separation.

15:20 – 15:40 Control abstraction by algebraic effect handlers
Taro Sekiyama (National Institute of Informatics)

Control flow is a crucial notion of running programs, determining the order in which instructions are executed. Because manipulating control flow in a well-behaved manner makes programs concise, intuitive, and structured, programming languages often provide constructs to operate control flow, such as exception handling. This talk will introduce algebraic effect handlers, a recently emerging construct for abstraction of control flow manipulation. Algebraic effect handlers generalize exception handling to allow the resumption of computation and have several applications including state, backtracking, generators, and probabilistic programming.

15:40 – 16:00 Brain computation as fast spiking neural inference in probabilistic programs
Vikash Mansinghka (MIT Probabilistic Computing Project)
16:00 – 17:00 Discussion
moderator:
Joi Ito (Chiba Institute of Technology)
Atsuhiro Takasu (National Institute of Informatics)

   Registration

Please contact workshop secretariat takasu[at]nii.ac.jp for webex url.

 

Copyright ©2022 National Institute of Informatics