Published Feb 28, 2022

#66 ALEXANDER MATTICK - [Unplugged / Community Edition]

Alexander Mattick and Tim Scarfe dive into the complexities of causal learning in machine learning, dissecting philosophical challenges and exploring neural network theories, while highlighting the importance of knowledge transfer for scientific advancement.
Episode Highlights
Machine Learning Street Talk (MLST) logo

Popular Clips

Episode Highlights

  • Knowledge Transfer

    The conversation between and explores the efficiency of knowledge transfer across different domains. Tim likens knowledge acquisition to the "curse of dimensionality," where information sharing grows exponentially with team size. However, he finds that understanding concepts like probability theory can simplify complex problems, akin to Tetris pieces fitting together 1. Alexander agrees, noting that many fields share underlying principles, allowing knowledge to be applied across domains. He emphasizes the importance of broad knowledge, not just for solving specific problems, but for gaining context and understanding 2.

       

    Concept Duality

    Tim introduces the idea of dualities in knowledge and control, drawing on Claude Shannon's insights about the past and future. He explains that scientists gain knowledge through controlled experiments, while engineers use knowledge to exert control over the future 3. Alexander adds that engineering advancements often lead to theoretical breakthroughs and vice versa, highlighting the interconnectedness of these domains. He recalls Richard Feynman's advice to scientists to keep multiple problems in mind, applying new solutions creatively to existing challenges 4.

       

    Interestingness

    The discussion shifts to the concept of "interestingness" in research, as and explore the idea of life as a search problem. Tim suggests that greatness might be just a door away, but finding it requires serendipity and subjective judgment 5. Alexander shares his view on using heuristics to focus on promising areas, while remaining open to unexpected discoveries. He stresses the importance of broad knowledge, as it prepares one for unforeseen opportunities and challenges 2.

Related Episodes