Machine Learning Street Talk (MLST) Prof. BERT DE VRIES - ON ACTIVE INFERENCE
92 snips
Nov 20, 2023 Bert de Vries, a professor at Eindhoven University, shares his expertise on intelligent autonomous agents. He discusses the principle of least action, its universal implications, and how it relates to optimizing energy in systems. Bert delves into active inference's role in ecosystems and engineering, particularly in adaptive technologies like hearing aids. He highlights the importance of probabilistic inference for the future of intelligence and explores challenges of implementing these concepts in real-world applications, emphasizing adaptability and innovation.
AI Snips
Chapters
Books
Transcript
Episode notes
Goal-Based Learning in Active Inference
- Active inference uses Bayesian surprise minimization and KL divergence for fidelity.
- Goal-based learning in active inference allows for applying specific pressures to achieve goals.
Bert de Vries's Introduction to Active Inference
- Bert de Vries's interest in active inference stemmed from a paper by Karl Friston.
- He realized its potential for creating self-organizing hearing aids that adapt to real-world situations.
Academic vs. Industrial Research
- Academic research often prioritizes interesting problems.
- Industrial research demands that problems be both interesting and important for practical application.
