Guest Editorial: Special Issue on Architectures and Design Methods for Neural Networks
- PDF / 149,372 Bytes
- 3 Pages / 595.276 x 790.866 pts Page_size
- 11 Downloads / 202 Views
Guest Editorial: Special Issue on Architectures and Design Methods for Neural Networks Ahmed Hemani 1 & Mohammed Shafique 2 & Kolin Paul 3 Received: 29 June 2020 / Revised: 29 June 2020 / Accepted: 29 June 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
The key strength and differentiation of humans as species has been their ability to harness external sources to supplement their physical and mental faculties. For a greater part of history, humans have harnessed combination of tools, natural forces like gravity and wind, and animal muscles to supplement their own weak muscles. A big breakthrough in supplementing weak biological muscles came about when we learned to mechanize generation of power and achieved affordable 2–3 orders greater power density compared to biological power. This transformed the society from being primarily agricultural to industrial. Parallel to the history of our attempts to supplement our physical strength has been a trend to supplement our mental strength. Development of writing, paper, printing, storage and transmission of information and computation have been some of the key steps in supplementing our mental strength with external tools and machines. In spite of this long and impressive history of progress in supplementing human abilities with external forces and tools, one aspect of humans has been considered as quintessentially human that cannot be mechanized or automated. This aspect is our ability to reason and create ideas – also known as intelligence. Human intelligence has been the key driver in our ability to supplement our physical and mental abilities. The question that has been discussed for more than a century is can
* Ahmed Hemani [email protected] Mohammed Shafique [email protected] Kolin Paul [email protected] 1
Department of EE, School of EECS, KTH (Royal Institute of Technology), Stockholm, Sweden
2
Institute of Computer Engineering, Vienna University of Technology, Vienna, Austria
3
Department of CSE, Indian Institute of Technology, Delhi, India
human intelligence create artificial intelligence to supplement it and perhaps exceed that for certain tasks? The jury is still out on this question and is likely to remain so for some time. At this stage, we do not have an accepted definition of what constitutes intelligence, or its crux in form of the cognition. Our earlier attempts at Artificial Intelligence (AI) based on what is known as symbolic computation did not succeed well. This is how Marvin Minsky summed up the status of symbolic AI in 1977: Our first foray into Artificial Intelligence was a program that did a credible job of solving problems in college calculus. Armed with that success, we tackled high school algebra; we found, to our surprise, that it was much harder. Attempts at grade school arithmetic, involving the concept of numbers, etc., provide problems of current research interest. An exploration of the child’s world of blocks proved insurmountable, except under the most rigidly constrained circumstances. It
Data Loading...
