Chapter
The problem of human intelligence
03:54 - 20:10 (16:16)
The problem of human intelligence has become a focus of research for scientists, including the need for deep learning techniques to have a lot of labeled examples. The question remains whether understanding the biological workings of the human brain is necessary to create advanced intelligence systems.
Clips
The problem of human intelligence became a real focus for scientists' research as the most interesting problem to solve.
03:54 - 10:32 (06:38)
Summary
The problem of human intelligence became a real focus for scientists' research as the most interesting problem to solve. It is questioned how far we can get in creating intelligence systems without understanding the biological basis of human intelligence.
ChapterThe problem of human intelligence
EpisodeTomaso Poggio: Brains, Minds, and Machines
PodcastLex Fridman Podcast
The challenge with deep learning today is the need for massive amounts of labeled data, which poses a problem for the human brain, as it does not learn through labeled data.
10:32 - 16:59 (06:27)
Summary
The challenge with deep learning today is the need for massive amounts of labeled data, which poses a problem for the human brain, as it does not learn through labeled data. Deep learning techniques were initially inspired by the layered hierarchical networks in the brain discovered in the 60s by Torsten Wiesel and David Hubel at Harvard.
ChapterThe problem of human intelligence
EpisodeTomaso Poggio: Brains, Minds, and Machines
PodcastLex Fridman Podcast
This podcast discusses the possibility of humans having a general learning machinery, hardwired through evolution with weak priors, as opposed to a specific learning system.
16:59 - 20:10 (03:10)
Summary
This podcast discusses the possibility of humans having a general learning machinery, hardwired through evolution with weak priors, as opposed to a specific learning system. It also explores the role of hardware and software in children's ability to learn new concepts.