Monday, 27 March 2017
DEEP LEARNING OR MACHINES AS THE NEXT ARCHITECTS OF KNOWLEDGE: Part B.
Image above (core photo): Deep neural network (taken at the Hub Science lecture on "What Is Deep Learning?" by Alexandros Karatzoglou: Machine Learning Scientific Director, Telefonica Research (Spain), and Omiros Papaspiliopoulos: ICREA Research Professor, Pompeu Fabra University (Spain).
Can computers learn themselves and fulfill complicated tasks? What does this mean for professionals of all kinds in the future? From Spotify (Discover Weekly) and Google (ImageNet) applications to television programs like House of Cards (2013) and from diagnostics/ radiology/ molecular biology to crowd-sourcing, translating, image and voice recognition analysis, DL is nested within Artificial Intelligence. The expansive adaptation of convolutional neural networks, (the core of DL algorithmic architecture) the last years in medical, educational and recreational services globally, far from reaching an executional error of less than 8% (the current plateau for humans), guarantees credible results in statistics (elections), auctions (digital marketing), cures (tumor prevention) and administration/reporting (micro-secretarial, voice translation and transcript tasks).
The tools of DL exist for the last decades but it is only recently that their vast modelization and application has been launched. The Big Data era, social infrastructures’ Big Models, the current computers /mobile phones with advanced graphic cards, the democratization of Media with new independent outlets, (therefore the large information diffusion), but also the world domination strategies of tech multinationals by software sharing and testing (:Theano, Python, Keras, etc) PLUS the international, economic crisis urging several guilds/vocations to cease activity- strengthen DL presence as a key to business. Thus includes design, crafts and architectural solutions- at least for now, to some extent.
Cognitive sciences, psychology and education remain subjects possibly for the next generation of DL testing. The ethical aspect together with the political correctness of DL-architected systems, still remain investigated and a field of debate (ex. The selection of candidates for a high-rank administrative position by DL software instead of a traditional HR office, will have to be based on equal opportunity criteria for gender, race and accessibility).
Can a parent, an artist, a writer, a guardian or a carer – be substituted by DL? As long as there will be humans to design, restore and require those systems- most probably. Otherwise what are humans really there for?-))
Please also check the BBC show Talking Business and its extensive, dedicated episodes to DL, hosted by Colm O'Regan and Karishma Vaswani:
NEXT POST AROUND THE 10TH OF APRIL: IT’S ALL ABOUT SALONE DEL MOBILE 2017!!! STAY TUNED!!!