- We suggest directions by which neuroscience could seek to refine and test these hypotheses.
- Two recent developments have emerged within machine learning that create an opportunity to connect these seemingly divergent perspectives.
- We hypothesize that (1) the brain optimizes cost functions, (2) these cost functions are diverse and differ across brain locations and over development, and (3) optimization operates within a pre-structured architecture matched to the computational problems posed by behavior.
- Cost functions and training procedures have become more complex and are varied across layers and over time.
- In machine learning artificial neural networks tend to eschew precisely designed codes, dynamics or circuits in favor of brute force optimization of a cost function, often using simple and relatively uniform initial architectures.
Read the full article, click here.
@neuroraf: “Integration of Deep Learning and Neuroscience by @AdamMarblestone @KordingLab & @DeepMindAI”
bioRxiv – the preprint server for biology, operated by Cold Spring Harbor Laboratory, a research and educational institution
Towards an integration of deep learning and neuroscience