Friday 19 March 2004

A.E.Brain - a short Eulogy

No, not for me. Nor for my son Andrew (thank God!). For my Uncle, Dr Alfred E. Brain, who I've just got the word died on the 13th, a few days before my birthday.

I'll always remember him as the provider of California Redwood play-blocks, when I was about Andrew's age. And a "Think-A-Dot" toy at age 7 or so, that first got me interested in computing. I couldn't fathom it out at the time, but 10 years later (and half a world away), when I was introduced to Flip-Flops and simple Arithmetic Logic Units at University, I instantly picked up the concepts.

ShakeyBut he was no mere provider of goodies from a distance. Though as he lived in Santa Cruz, California, while I was in Berkshire, England, we didn't get to see him much. In the 60's, he was busy at Stanford Research International, working on Artificial Intelligence and Robotics.

Before then, he'd been working on Perceptrons, in particular MINOS I, widely regarded as the first successful attempt at rudimentary AI (Artificial Intelligence).

This work was discontinued almost entirely due to one very influential paper by Marvin Minsky (one of the Gods of Computer Science) in 1969, which showed that single-layer Perceptrons had fundamental limitations. Alas, Minsky didn't realise that multi-layer Perceptrons - Neural Nets - had no such limitations. The funding dried up, and it wasn't until the 90's that research on Neural Nets resumed, research that continues today. In fact, virtually all AI is now done using Neural Nets. We lost 20 years.

I can't give many details about my Uncle's work in the 40's and 50's, as he couldn't talk about it. During World War II, he was engaged in Electronic Warfare research and development at the University of Manchester, and that work's still classified. It may have been on the Monika tail-warning radar, but given his later career, was more probably work on ULTRA. After the war, he emigrated to the USA, and was involved in research on automated recognition of interesting things on recon photographs, which led to his work on Perceptrons. The US had a number of recon programs before the U-2 AQUATONE spyplane, usually involving balloon-borne cameras ( e.g. the WS-119L GENETRIX ) that took literally millions of metres of film over semi-random areas of the USSR. To manually go through every frame, looking for items of interest was infeasible, it had to be automated.

As an aside, the earlier Project MOGUL caused the most famous UFO incident ever recorded - the "Roswell Flying Saucer Crash". Though there's no convincing some people.

In the 70's, Uncle Ted ( as I knew him ) was still working on visual recognition, co-authoring a famous paper on the subject with Nizam and Duda ( Nizam77). And in the 80's, after retiring at 65 ( a near-fatal heart attack followed by a triple bypass encouraged him to slow down a bit), he returned to England and married his childhood sweetheart!

Like me, he had a few problems due to his name, and field of study. At least one article was rejected by the Scientific American as they thought a paper on An Electronic Brain by A.E.Brain was a hoax...

I last saw him in 1990 - when he met Carmel for the first time. His house in Market Raizen was full of clocks, as he'd got more than a passing interest in Horology. (He was moderately well known known in Horological circles for some papers on the history of the Waltham company). Also there was an electronic dog (which barked when anyone went near the front door), and his garden was filled with electronic Mole-Scarers.

His mind was as sharp as ever to the last, but his body let him down. Increasingly frail over the last few years, he still managed to write the odd article or two, the last one on Global Warming. Had his body not let him down, no doubt he would be blogging. But it was not to be.

If there is anything to this Afterlife and Religious bit, then no doubt he and my dad (his younger brother) are out there "Drowning Worms" or whatever the celestial equivalent is. Though I rather think they might be giving the Almighty a bit of a hand in the Engineering of the Universe.

In any event, Andrew's going to hear a lot more about this other A.E.Brain. He could do a lot worse than to be like him.



No comments: