Machine Learning Street Talk (MLST) Capsule Networks and Education Targets
Sep 29, 2020
In this engaging discussion, guest Alex Stenlake, a regular contributor on AI and capsule networks, explores the education chapter of Kenneth Stanley's book, questioning the value of objective-driven incentives and their impact on creativity. The conversation shifts to capsule networks, detailing their capabilities in feature detection and spatial representation. They also tackle the challenges in optimizing deep learning algorithms, the potential of quantum computing, and the need for innovative thinking versus rigid educational structures.
AI Snips
Chapters
Books
Transcript
Episode notes
Objectives and Innovation
- Objectives can stifle innovation and create strange side effects when directly optimized.
- Educational objectives should meaningfully correlate with societal goals, like skills needed for specific jobs.
Cobra Effect in India
- The British government's attempt to exterminate snakes in India by offering bounties backfired.
- Citizens began breeding cobras for profit, increasing the snake population.
The Value of Exploration
- Exploring without a clear objective is often seen as flawed, even by successful innovators like Steve Jobs.
- True innovation requires a willingness to diverge from established metrics and explore uncharted territories.
