Gary Marcus

Rank 25 of 47
|
Score 40
1/11/2025, 6:57:21 PM
X
In reply to:
davinci
@basedneoleo
·
26d
@GaryMarcus that's what the depth in deep learning is for...
Gary Marcus
@GaryMarcus
·
26d
@basedneoleo to get to the right level of abstraction. see my 2001 book The Algebraic Mind for a long argument about this.
davinci
@basedneoleo
·
26d
@GaryMarcus why?
Gary Marcus
@GaryMarcus
·
26d
@basedneoleo i am all for better incorporating causality, though i think you will need symbols as part of the package to do it right.
davinci
@basedneoleo
·
26d
@GaryMarcus okay sure i'll rephrase this without the "vibes".

principled methods for tackling the fragility under distribution shifts, and consequently the poor OOD generalization of neural networks, build on top of the causality literature, particularly causal representation learning. the…
Gary Marcus
@GaryMarcus
·
26d
@basedneoleo

The statement is part of a technical discussion on deep learning and its limitations, specifically addressing the issue of abstraction and causality in neural networks. It references a previous argument made by the speaker in 2001, suggesting a principled reason why deep learning hasn't solved the problem discussed.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not use harmful language and maintains a professional tone, adhering to the principle of doing no harm with words. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement aims to promote understanding by referencing a principled reason and a previous work, although it could be more explicit in explaining the reasoning to enhance empathy and compassion. [+1]
  3. Principle 4:
    I will engage in constructive criticism and dialogue with those in disagreement and will not engage in personal attacks or ad hominem arguments.
    The statement engages in constructive criticism by providing a reasoned argument against the effectiveness of deep learning in solving the problem, without resorting to personal attacks. [+2]