Brain-Like Object Recognition with High-Performing Shallow Recurrent ANNs

By Jonas Kubilius, Martin Schrimpf, Ha Hong, Najib J. Majaj, Rishi Rajalingham, Elias B. Issa, Kohitij Kar, Pouya Bashivan, Jonathan Prescott-Roy, Kailyn Schmidt and 4 more. Neural Information Processing Systems, 2019

333 citations recorded by the source corpus.

Deep convolutional artificial neural networks (ANNs) are the leading class of candidate models of the mechanisms of visual processing in the primate ventral stream. While initially inspired by brain anatomy, over the past years, these ANNs have evolved from a simple eight-layer architecture in AlexNet to extremely deep and branching architectures, demonstrating increasingly better object categorization performance, yet bringing into question how brain-like they still are. In particular, typical deep models from the machine learning community are often hard to map onto the brain's anatomy due to their vast number of layers and missing biologically-important connections, such as recurrence. Here we demonstrate that better anatomical alignment to the brain and high performance on machine learning as well as neuroscience measures do not have to be in contradiction. We developed CORnet-S, a shallow ANN with four anatomically mapped areas and recurrent connectivity, guided by Brain-Score, a new large-scale composite of neural and behavioral benchmarks for quantifying the functional fidelity of models of the primate ventral visual stream. Despite being significantly shallower than most models, CORnet-S is the top model on Brain-Score and outperforms similarly compact models on ImageNet. Moreover, our extensive analyses of CORnet-S circuitry variants reveal that recurrence is the main predictive factor of both Brain-Score and ImageNet top-1 performance. Finally, we report that the temporal evolution of the CORnet-S "IT" neural population resembles the actual monkey IT population dynamics. Taken together, these results establish CORnet-S, a compact, recurrent ANN, as the current best model of the primate ventral visual stream.

Subfields: Artificial Intelligence, Cognitive Neuroscience, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/146d40346ed686f54087386f66666adaf6a6efaf

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Cognitive Neuroscience

In Computer Vision and Pattern Recognition

Citations in the corpus

Integrative Benchmarking to Advance Neurally Mechanistic Models of Human Intelligence. (2020, background): “Our ﬁrsthintofsuccessofthisstrategyiswiththemodelCORnet,whichwasbuiltwithguidancefromneuroscience(recurrence),andbecauseitsBrain-Scoreishigherthanpriormodels,thatmodelbuildingeffortisthusrewarded.”
Unsupervised neural network models of the ventral visual stream (2020, background): “Recent work has begun to integrate into neural networks analogs of the recurrences and long-range feedbacks that have been ubiquitously observed throughout the visual system, toward better modeling neural dynamics (74, 75).”
Beyond accuracy: quantifying trial-by-trial behaviour of CNNs and humans by measuring error consistency (2020, methodology): “ Networks. Human responses were compared against classiﬁcation decisions of all available CNN models from the PyTorch model zoo (for torchvision version 0.2.2) and against a recurrent model, CORnet-S [38]. All CNNs were trained on ImageNet. Details here: S.4. 3 Results If two perceptual systems or d”
Synaptic Plasticity Dynamics for Deep Continuous Local Learning (DECOLLE) (2018, methodology): “In shallow convolutional neural networks, it is common to use larger kernel sizes (LeCun et al., 1998; Kubilius et al., 2019) ( Table 2 ).”
The neural architecture of language: Integrative modeling converges on predictive processing (2020, methodology): “In models of vision, for instance, attempts are made to map ANN layers and neurons onto cortical regions (3) and subregions (148).”
Convolutional Neural Networks as a Model of the Visual System: Past, Present, and Future (2020, background): “These connections can also help make the CNN representations a better match to neural data, particularly for challenging images and at later time points in the response (Kar et al., 2019; Kubilius et al., 2019; Rajaei, Mohsenzadeh, Ebrahimpour, & KhalighRazavi, 2019; Shi, Wen, Zhang, Han, & Liu, 201”
Partial success in closing the gap between human and machine vision (2021, neutral): “This includes comparing against neural data [84, 85], contrasting Gestalt effects [e.g. 86], object similarity judgments [87], or mid-level properties [61] and is of course not limited to studying object recognition.”