Why All Models Learn the Same Thing with Phillip Isola (MIT)
MIT professor Phillip Isola discusses what makes learned representations useful and why independently trained models can develop similar internal structure. He explores the platonic representation hypothesis, local clustering versus global geometry, and the neural thickets account of why pretrained networks can adapt readily to downstream tasks. Other topics include language models as world models, recurrent architectures, biological comparisons, and studying autonomous language-model agents as artificial life. Hosted by Ravid Shwartz Ziv and Allen Roush; watch the full research conversation on YouTube.
Learning and prediction in artificial deep neural networks: scaling, data manifolds, and universality
Developing scientifically-grounded theories for representation learning and generalization in artificial deep neural networks remains a grand challenge of fundamental interest to theoretical neuroscience and machine learning. I will discuss our work on one facet of this challenge — namely understanding generalization or “scaling laws” in learned neural networks as a function of basic control variables. I’ll discuss a taxonomy we develop that classifies different regimes of scaling behavior. We identify regimes where generalization exhibits universal scaling behavior and others where it can be traced back to properties of the data and neural architecture. The theoretical analysis is enabled by leveraging exactly solvable models of deep neural networks that arise naturally in the limit of large hidden layers. Along the way, I’ll also discuss our work on these theoretical models, which have been a useful starting point for theoretical descriptions of neural network dynamics. Finally, I’ll discuss our findings connecting generalization in neural networks to properties of the learned data manifold. I’ll close by discussing future directions and new hypotheses that emerge from our findings Presented in the van Vreeswijk Theoretical Neuroscience Seminar series (formerly WWTNS) on 2024-06-19. Recording duration: 00:46:53.
Deriving local synaptic learning rules for efficient representations in networks of spiking neurons
How can neural networks learn to efficiently represent complex and high-dimensional inputs via local plasticity mechanisms? Classical models of representation learning assume that input weights are learned via pairwise Hebbian-like plasticity. Here, we show that pairwise Hebbian-like plasticity only works under specific requirements on neural dynamics and input statistics. To overcome these limitations, we derive from first principles a learning scheme based on voltage-dependent synaptic plasticity rules. Here, inhibition learns to locally balance excitatory input in individual dendritic compartments, and thereby can modulate excitatory synaptic plasticity to learn efficient representations. We demonstrate in simulations that this learning scheme works robustly even for complex, high-dimensional and correlated inputs. It also works in the presence of inhibitory transmission delays, where Hebbian-like plasticity typically fails. Our results draw a direct connection between dendritic excitatory-inhibitory balance and voltage-dependent synaptic plasticity as observed in vivo, and suggest that both are crucial for representation learning.
One Instructional Sequence Fits all? A Conceptual Analysis of the Applicability of Concreteness Fading
According to the concreteness fading approach, instruction should start with concrete representations and progress stepwise to representations that are more idealized. Various researchers have suggested that concreteness fading is a broadly applicable instructional approach. In this talk, we conceptually analyze examples of concreteness fading in mathematics and various science domains. In this analysis, we draw on theories of analogical and relational reasoning and on the literature about learning with multiple representations. Furthermore, we report on an experimental study in which we employed concreteness fading in advanced physics education. The results of the conceptual analysis and the experimental study indicate that concreteness fading may not be as generalizable as has been suggested. The reasons for this limited generalizability are twofold. First, the types of representations and the relations between them differ across different domains. Second, the instructional goals between domains and the subsequent roles of the representations vary.
Cross Domain Generalisation in Humans and Machines
Recent advances in deep learning have produced models that far outstrip human performance in a number of domains. However, where machine learning approaches still fall far short of human-level performance is in the capacity to transfer knowledge across domains. While a human learner will happily apply knowledge acquired in one domain (e.g., mathematics) to a different domain (e.g., cooking; a vinaigrette is really just a ratio between edible fat and acid), machine learning models still struggle profoundly at such tasks. I will present a case that human intelligence might be (at least partially) usefully characterised by our ability to transfer knowledge widely, and a framework that we have developed for learning representations that support such transfer. The model is compared to current machine learning approaches.
Unsupervised deep learning identifies semantic disentanglement in single inferotemporal neurons
Irina is a research scientist at DeepMind, where she works in the Froniers team. Her work aims to bring together insights from the fields of neuroscience and physics to advance general artificial intelligence through improved representation learning. Before joining DeepMind, Irina was a British Psychological Society Undergraduate Award winner for her achievements as an undergraduate student in Experimental Psychology at Westminster University, followed by a DPhil at the Oxford Centre for Computational Neuroscience and Artificial Intelligence, where she focused on understanding the computational principles underlying speech processing in the auditory brain. During her DPhil, Irina also worked on developing poker AI, applying machine learning in the finance sector, and working on speech recognition at Google Research."" https://arxiv.org/pdf/2006.14304.pdf