Thomas Serre's guest appearances

Thomas Serre

Brain Inspired

James FitzGerald

Jessica Thompson

David Rosen

Ryan Poplin

Blake Richards

Terrence Sejnowski

David Ferrucci

Raia Hadsell

Melanie Mitchell

Matt Smith

Alison Gopnik

Brian Christian

Robert Wilson

Anil Seth

Jeff Hawkins

OwlTail

Thomas and I discuss the role of recurrence in visual cognition: how brains somehow excel with so few “layers” compared to deep nets, how feedback recurrence can underlie visual reasoning, how LSTM gate-like processing could explain the function of canonical cortical microcircuits, the current limitations of deep learning networks like adversarial examples, and a bit of history in modeling our hierarchical visual system, including his work with the HMAX model and interacting with the deep learning folks as convolutional neural networks were being developed. Show Notes: Visit the Serre Lab website. Follow Thomas on twitter: @tserre.Good reviews that references all the work we discussed, including the HMAX model: Beyond the feedforward sweep: feedback computations in the visual cortex. Deep learning: the good, the bad and the ugly. Papers about the topics we discuss: Complementary Surrounds Explain Diverse Contextual Phenomena Across Visual Modalities. Recurrent neural circuits for contour detection.Learning long-range spatial dependencies with horizontal gated-recurrent units.

<p>Thomas and I discuss the role of recurrence in visual cognition: how brains somehow excel with so few “layers” compared to deep nets, how feedback recurrence can underlie visual reasoning, how LSTM gate-like processing could explain the function of canonical cortical microcircuits, the current limitations of deep learning networks like adversarial examples, and a bit of history in modeling our hierarchical visual system, including his work with the HMAX model and interacting with the deep learning folks as convolutional neural networks were being developed.</p> <p>Show Notes:</p> <ul><li>Visit the <a href="https://serre-lab.clps.brown.edu/" rel="nofollow" target="_blank">Serre Lab</a> website. </li><li>Follow Thomas on twitter: <a href="https://twitter.com/tserre" rel="nofollow" target="_blank">@tserre</a>.</li><li>Good reviews that references all the work we discussed, including the HMAX model: <ul><li><a href="https://nyaspubs.onlinelibrary.wiley.com/doi/abs/10.1111/nyas.14320" rel="nofollow" target="_blank">Beyond the feedforward sweep: feedback computations in the visual cortex</a>. </li><li><a href="https://www.annualreviews.org/doi/full/10.1146/annurev-vision-091718-014951" rel="nofollow" target="_blank">Deep learning: the good, the bad and the ugly</a>. </li></ul></li><li>Papers about the topics we discuss: <ul><li><a href="https://serre-lab.clps.brown.edu/wp-content/uploads/2019/12/Serre_092018.pdf" rel="nofollow" target="_blank">Complementary Surrounds Explain Diverse Contextual Phenomena Across Visual Modalities</a>. </li><li><a href="https://openreview.net/forum?id=H1gB4RVKvB" rel="nofollow" target="_blank">Recurrent neural circuits for contour detection</a>.</li><li><a href="https://arxiv.org/abs/1805.08315" rel="nofollow" target="_blank">Learning long-range spatial dependencies with horizontal gated-recurrent units.</a></li></ul></li></ul>

BI 065 Thomas Serre: How Recurrence Helps Vision