Tag: computer vision
-
Congratulations to Spandan Madan on his NeuroAI 2024 work!
10/10/2024 Improving out-of-distribution generalization by mimicking the human visual diet Spandan’s paper was accepted at NeuroAI @ NeurIPS 2024 In this work Spandan shows that a careful account of the visual input statistics that humans encounter during learning can help improve machine learning algorithms. Read more
-
Kreiman lab posters at SFN 2024
october 5 — OCTOBER 9, 2024 Presentations by the Kreiman lab at the Society for Neuroscience Annual Meeting 2024 Oct 5 2024. 3:30-3:45 Prefrontal and Medial Temporal Neurons Encode Ordinal Information of Event Sequences in Humans. Presented by Jie Zheng. Oct 7 2024, 8:00-13:00 Syntactic-feature encoding of language in the human brain. Presented by Pranav… Read more
-
Prof. Kreiman gives a lecture at UC Berkeley
09/18/2024 Successes and challenges in computational models of vision We now have powerful computer vision algorithms that can segment scenes, label objects, and recognize actions. It is tempting to use these algorithms as models of visual processing in biological brains. I will provide an overview of some of the successes in using neural network models to partially describe visual behavior and predict neuronal responses in the visual system of monkeys and humans, focusing on the ventral stream associated with object recognition.… Read more
-
The next 200 milliseconds
08/08/2024 Prof. Kreiman gives a lecture on the dynamics of cognition at the Brains, Minds and Machines This lecture introduces a discussion of the complex recurrent and top-down signals that interact with sensory processes to interpret visual scenes, including contextual reasoning, eye movements and visual search, and pattern completion. Read more
-
The limits of computer vision
05/15/2024 Harvard Medicine News Work from the Kreiman laboratory about the successes and limitations of computer vision algorithms was featured in this article by Molly McDonough. Read more
-
Current models of visual cortex fail to generalize
05/01/2024 Spandan Madan gives a talk at the Center for Brain Science Neurolunch Spandan scrutinizes the ability of models of visual cortex to explain neuronal responses and finds a poor ability to extrapolate to novel stimuli, forcing us to revisit whether such models capture the critical properties of vision. Read more
-
Will Xiao on visual cortex activity during free viewing
04/29/2024 Feature-selective responses in macaque visual cortex follow eye movements during natural vision How do visual cortex neurons respond during free viewing conditions? Will Xiao’s tour-de-force work answering this question was published in Nature Neuroscience. Read more
-
A Clue Into How We See a Stable World Despite Moving Eyes
04/29/2024 Harvard Medicine News and Research We move our eyes about three times per second, yet the world seems perfectly stable. Will Xiao’s tour-de-force investigations provide the most extensive computational and neurophysiological description of visual cortical responses during free viewing. See Xiao’s publication here. Read more