Monday, April 19, 2021

Тошко 2.075 - безплатен синтезатор на реч, говореща програма | Toshko 2.075 Bulgarian TTS engine

Поправка на малък бъг, свързан с работата при липса на Интернет връзка.
Виж и изтегли от: https://github.com/Twenkid/Toshko_2

Saturday, April 17, 2021

Is the new theory of the "Thousand Brains" by Jeff Hawkins a new one? - and an announcement about the yet unpublished book How an 18-year OId Prodigy Researcher Preceded MIT and Stanford by 15 years in AGI

The second book by Hawkins was published in March: A Thousand Brains: A New Theory of Intelligence Hardcover – March 2, 2021 by Jeff Hawkins (Author), Richard Dawkins (Foreword) https://www.amazon.com/Thousand-Brains-New-Theory-Intelligence/dp/1541675819

It might really be new regarding the neocortex, but I didn't find it so new reg. AGI and epistemology. In brief, all these directions, working with universal/general "reference frames", 3D-space, search for coordinates and their relations follow directly and obviously from the sensorimotor school of thought and they have been expressed in many places, for example I'd say even in "Five basic principles of developmental robotics" by Stoychev (see the discussion in the blog), my older than that theory and publications and Hawkins' own earler book "On Intelligence" (2004). 

 For a more extensive review of why these ideas do not seem so new see my upcoming book "How an 18-year old prodigy researcher preceded MIT and Stanford by 15 years in AGI", originally in Bulgarian: "Как 18-годишен информатик от Пловдив изпревари MIT и Станфорд с 15 години във всеобщия изкуствен интелект, или Универсален изкуствен разум: Artificial General Intelligence".

It is a call for collaborators, cofounders, researchers, supporters, sponsors for the creation of an interdisciplinary institute about which I wrote in my teenage essay 18 years ago in 2003: "How would I invest one million Euro to achieve the best progress of my country" (see in the links) - MIT and Stanford created interdisciplinary institutes with somewhat similar headlines in 2018 (MIT claiming one BILLION $ investments), the famous Bulgarian AI researcher Martin Vechev, leader of the startup "DeepCode" for code synthesis, together with Bulgarian Academy of Science are also about to create an academic AI institute and infrastructure in Bulgaria.

My book is a historical documentary collection of various works, publications and comments, full or excerpts, mainly mine and references to other contemporary researchers' and other articles about the topics, announcements about events etc., presented in chronological order, with added additional notes. 

The work was in a somewhat "complete" state since August 2020, but I've been hesitating how exactly to present it. It seems unlikely to get it printed commercially: it's huge, in current formatting 1028 pages A4 in the full version. It could be condensed a bit due to some deliberate repetitions (for finding some texts in a shorter and longer chronology). However I don't see interest for now.

The work on the book was on hold, until last month when I added a lot of material and editions, including notes on that new Hawkins' book and other discoveries of "new" ideas in AGI which happen to be repeating ones expressed in my early 2000s works, such as one 2021 Google Brain researcher's generalisation about what general intelligence is about - see in the book; I may post an article in the blog as well.