Chapter
Solving the Traveling Salesman Problem with Markus Hutter's Universal Algorithm
06:51 - 19:16 (12:25)
Markus Hutter's universal algorithm can solve the Traveling Salesman Problem in n to the 5 steps, plus O of 1, with the constant overhead for proof search guaranteeing optimality, while the more practical recurrent neural networks and local search techniques are used for solving small problems like speech recognition and machine translation.
Clips
The development of universal problem solvers, like the Gödel machine and Markus Hutter's method, come with the overheads of proof search, ensuring that the solutions found are optimal.
06:51 - 11:31 (04:39)
Summary
The development of universal problem solvers, like the Gödel machine and Markus Hutter's method, come with the overheads of proof search, ensuring that the solutions found are optimal. However, non-universal techniques like recurrent neural networks and local search are much more practical for solving smaller problems.
ChapterSolving the Traveling Salesman Problem with Markus Hutter's Universal Algorithm
EpisodeJuergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs
PodcastLex Fridman Podcast
This podcast discusses the use of local search gradient descent to find a program running on recurrent neural networks (RNN) for solving problems such as speech recognition or machine translation, despite the practical problem solvers being a different kind of AI.
11:31 - 15:34 (04:03)
Summary
This podcast discusses the use of local search gradient descent to find a program running on recurrent neural networks (RNN) for solving problems such as speech recognition or machine translation, despite the practical problem solvers being a different kind of AI.
ChapterSolving the Traveling Salesman Problem with Markus Hutter's Universal Algorithm
EpisodeJuergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs
PodcastLex Fridman Podcast
The possibility of creating human-level intelligence in AI is dependent on building on existing knowledge, such as fundamental mathematical operations, but the apparent randomness of some events in the universe poses a challenge for a shortcut.
15:34 - 19:16 (03:42)
Summary
The possibility of creating human-level intelligence in AI is dependent on building on existing knowledge, such as fundamental mathematical operations, but the apparent randomness of some events in the universe poses a challenge for a shortcut.