Supervised learning

…as an interface, training with supervised learning, rejection sampling and reinforcement learning, and coordinating tools and sub-agents. It considers reliability and the progression from background monitoring to autonomous web tasks.

…such models can be learned, both using supervised learning and using implicit feedback as an agent interacts with the environment, the latter being more plausible in biological agents. We also discuss two ways such models can be used. Firs

…signal underlying this normative learning process remains elusive. Here, we test two well-established and biologically plausible theories---supervised learning (SL) and reinforcement learning (RL)---that could explain how neural circuits develop the capacity for skilled movements. We trained recurrent neural networks to control a biomechanical model of

Task acquisition in the brain relies on both reward-based and error-based learning (RBL, EBL), yet how the brain brings these two learning signals together remains elusive. Here, we build on the traditional view that dopamine (DA) projections within the basal ganglia (BG) drive RBL via reinforcement learning, while the cerebellum (CB) supports EBL via supervised learning. In doing so, we introduce a novel learning framework that combines DA-driven RBL with CB-driven EBL. Specifically, we reformulate the traditional RBL and EBL synaptic weight updates in terms of a shared ‘action-gradient’ spac

…Existing models have focused on supervised learning, in which the auditory system is calibrated by exactly matching feedback signals from the precise visual system. But such cross-sensory calibration is not always feasible. It is unclear how to calibrate the auditory map when

…studies focus on the end-result of the learning process by measuring and comparing the similarities in the representations of object categories once they have been formed. However, the process of how these representations emerge—that is, the behavioral changes and intermediate stages observed during the acquisition—is less often directly and empirically compared. In this talk, I'm going to report a detailed investigation of the learning dy

…Its differentiability allows for a supervised learning approach usin

…explores the application of deep learning techniques for compressing extracellular neural data, leveraging the hierarchical representations learned by convolutional neural networks to achieve efficient compression without com

Flying insects can process multiple visual features in parallel neural circuits and generate an appropriate action. Neural processing of singly presented visual patterns has been studied intensively in Drosophila for the past few decades. How do parallel visual circuits responding to different features, presented in a single visual scene, are integrated to control a shared motor circuit? An influential theory proposed for combining multiple sensorimotor circuits is an efference copy mechanism, in which an intended action offsets other sensory circuits to prevent them from responding to reaffe

Animals naturally display a wide range of behaviours in varying contexts. Not only do their actions depend on their internal states, but also how their early sensory cortex processes information. Here, we propose that contextual top-down afferents adapt the receptive fields of sensory neurons to accommodate varying task demands. Although this view is widely supported by connectomic and electrophysiological evidence, a mechanistic and, in particular, a normative understanding is still lacking. We investigate how task-dependent top-down signals can reshape the functional mapping of sensory proce

…on deep neural networks trained using supervised learning, in tasks such as image classification. However, there is another area of recent AI work which has so far received less attention from neuroscientists, but which may have more profound neuroscientific implications: Deep reinforcement learning. Deep RL offers a rich framework for studying the interplay among learning, representation