Alex Yakovlev: “Data-Driven Computing”
Join guest professor Alex Yakovlev as he explores the potential for designing a data-driven computer based on the concept of an ensemble of learning automata.
This is an online-only event.
See description for details.
Traditional computing is based on the Von Neumann’s principle of a stored program. This paradigm requires a central processing unit (CPU) to access memory at each step in the algorithm. The memory becomes a bottleneck as it stands on the critical path of the compute actions. Data-driven computing is based on automatic synthesis of the computing configuration “on the fly” (as a process analogous to compilation) from the data collected by human or another computing system and presented in the form of the information about the object to be classified (e.g., recognized) labelled with classification labels, called labelled data mapping (LDM).
This talk will explore potential for designing a data-driven computer based on the concept of an ensemble of learning automata. These learning automata can be trained to perform configuration of computing resources based on LDM. The core of the new machine will be a special type of multicast memory with in-memory computing capability and efficient accessing for reading their states and their state modification. The future of data-driven computing can be projected at many levels of abstraction and with different implementation technologies.
About Alex Yakovlev
Alex Yakovlev is an international pioneer of low-power asynchronous circuit design and automation, for which he was elected to Fellow of IEEE, RAeng and IET. He is Professor of Computing Systems Design at the School of Engineering, Newcastle University, where he has been working since 1991. He received DSc from Newcastle University in 2006, and PhD from St. Petersburg Electrical Engineering Institute (USSR) in 1982, both in the field of asynchronous systems. At Newcastle, since 2000 he is Head of the Microsystems Group and Founder of the Asynchronous Systems Lab, with over 60 PhD alumni. His team is well-known for its contributions in designing asynchronous circuits, concurrent systems, Petri nets, metastability and synchronizers. His most recent work is in the fields of electromagnetic computing and circuit design for machine learning based on Tsetlin automata.
The lecture series on research talks by the guest professors of the TU Wien Informatics Doctoral School can also be credited as an elective course for students of master programs of computer science: 195.072 Current Trends in Computer Science.