What neural machinery is needed for language acquisition (Virtual Talk)
Talkat Understanding Higher-Level Intelligence from AI, Psychology, and Neuroscience Perspectives · 2024-06-24..06-28·
summary
A minimal neural model called NEMO postulates cortical assemblies, discrete timesteps, random sparse synapses, Hebbian updates and long-range inhibitory populations; it simulates about 100 million neurons (areas with n≈1e6, k≈1e3 active) and learns from a few hundred sentences grounded to sensory/motor activations. The system implements a parser (ACL paper), forms lexicons separating nouns/verbs, produces dependency trees and generates correct captions for novel scenes — a working proof of concept that handles concrete language but requires extra mechanisms for abstract and social words.
AI-written summary · about
chapters (9)
- 0:00Simons Institute intro
- 2:20Workshop goals
- 9:13Neural basis of language
- 14:29NEMO brain model
- 29:36Language parsing in NEMO
- 33:43Biologically plausible acquisition
- 47:39Abstract and social language challenges
- 55:08Future directions
- 1:06:27Q&A
related talks
40mIs my Large Language Model a Strange Loop?
01h 04mA Very Small Language Model
01h 11mThe Move Toward AGI: Why Large Language Models Surprised Almost Everyone... | Theoretically Speaking
46mWhen will we have a Rosie of our own?
01h 01mBetween the Layers– Interpreting Large Language Models
20mMultimodal language models
40mKeynote: Open-source Multimodal AI in the Wild
30mLarge Language Models and Prompt Engineering for Developers