Unsupervised approaches to learning in neural networks are of substantial interest for furthering artificial intelligence, both because they would enable the training of networks without the need for large numbers of expensive annotations, and because they would be better models of the kind of general-purpose learning deployed by humans. However, unsupervised networks have long lagged behind the performance of their supervised counterparts, especially in the domain of large-scale visual recognition. Recent developments in training deep convolutional embeddings to maximize non-parametric instance separation and clustering objectives have shown promise in closing this gap. Here, we describe a method that trains an embedding function to maximize a metric of local aggregation, causing similar data instances to move together in the embedding space, while allowing dissimilar instances to separate. This aggregation metric is dynamic, allowing soft clusters of different scales to emerge. We evaluate our procedure on several large-scale visual recognition datasets, achieving state-of-the-art unsupervised transfer learning performance on object recognition in ImageNet, scene recognition in Places 205, and object detection in PASCAL VOC.

Deep neural networks currently provide the best quantitative models of the response patterns of neurons throughout the primate ventral visual stream. However, such networks have remained implausible as a model of the development of the ventral stream, in part because they are trained with supervised methods requiring many more labels than are accessible to infants during development. Here, we report that recent rapid progress in unsupervised learning has largely closed this gap. We find that neural network models learned with deep unsupervised contrastive embedding methods achieve neural prediction accuracy in multiple ventral visual cortical areas that equals or exceeds that of models derived using today’s best supervised methods and that the mapping of these neural network models’ hidden layers is neuroanatomically consistent across the ventral stream. Strikingly, we find that these methods produce brain-like representations even when trained solely with real human child developmental data collected from head-mounted cameras, despite the fact that these datasets are noisy and limited. We also find that semisupervised deep contrastive embeddings can leverage small numbers of labeled examples to produce representations with substantially improved error-pattern consistency to human behavior. Taken together, these results illustrate a use of unsupervised learning to provide a quantitative model of a multiarea cortical brain system and present a strong candidate for a biologically plausible computational theory of primate sensory learning.

Deep neural networks currently provide the best quantitative models of the response patterns of neurons throughout the primate ventral visual stream. However, such networks have remained implausible as a model of the development of the ventral stream, in part because they are trained with supervised methods requiring many more labels than are accessible to infants during development. Here, we report that recent rapid progress in unsupervised learning has largely closed this gap. We find that neural network models learned with deep unsupervised contrastive embedding methods achieve neural prediction accuracy in multiple ventral visual cortical areas that equals or exceeds that of models derived using today’s best supervised methods, and that the mapping of these neural network models’ hidden layers is neuroanatomically consistent across the ventral stream. Moreover, we find that these methods produce brain-like representations even when trained on noisy and limited data measured from real children’s developmental experience. We also find that semi-supervised deep contrastive embeddings can leverage small numbers of labelled examples to produce representations with substantially improved error-pattern consistency to human behavior. Taken together, these results suggest that deep contrastive embedding objectives may be a biologically-plausible computational theory of primate visual development.

No abstract

No abstract

Visual information in the visual cortex is processed in a hierarchical manner. Recent studies show that higher visual areas, such as V2, V3, and V4, respond more vigorously to images with naturalistic higher-order statistics than to images lacking them. This property is a functional signature of higher areas, as it is much weaker or even absent in the primary visual cortex (V1). However, the mechanism underlying this signature remains elusive. We studied this problem using computational models. In several typical hierarchical visual models including the AlexNet, VggNet, and SHMAX, this signature was found to be prominent in higher layers but much weaker in lower layers. By changing both the model structure and experimental settings, we found that the signature strongly correlated with sparse firing of units in higher layers but not with any other factors, including model structure, training algorithm (supervised or unsupervised), receptive field size, and property of training stimuli. The results suggest an important role of sparse neuronal activity underlying this special feature of higher visual areas.

Task-optimized deep convolutional neural networks are the most quantitatively accurate models of the primate ventral visual stream. However, such networks are implausible as a model of the mouse visual system because mouse visual cortex has a known shallower hierarchy and the supervised objectives these networks are typically trained with are likely neither ethologically relevant in content nor in quantity. Here we develop shallow network architectures that are more consistent with anatomical and physiological studies of mouse visual cortex than current models. We demonstrate that hierarchically shallow architectures trained using contrastive objective functions applied to visual-acuity-adapted images achieve neural prediction performance that exceed those of the same architectures trained in a supervised manner and result in the most quantitatively accurate models of the mouse visual system. Moreover, these models' neural predictivity significantly surpasses those of supervised, deep architectures that are known to correspond well to the primate ventral visual stream. Finally, we derive a novel measure of inter-animal consistency, and show that the best models closely match this quantity across visual areas. Taken together, our results suggest that contrastive objectives operating on shallow architectures with ethologically-motivated image transformations may be a biologically-plausible computational theory of visual coding in mice.

Because of the rich dynamical structure of videos and their ubiquity in everyday life, it is a natural idea that video data could serve as a powerful unsupervised learning signal for training visual representations in deep neural networks. However, instantiating this idea, especially at large scale, has remained a significant artificial intelligence challenge. Here we present the Video Instance Embedding (VIE) framework, which extends powerful recent unsupervised loss functions for learning deep nonlinear embeddings to multi-stream temporal processing architectures on large-scale video datasets. We show that VIE-trained networks substantially advance the state of the art in unsupervised learning from video datastreams, both for action recognition in the Kinetics dataset, and object recognition in the ImageNet dataset. We show that a hybrid model with both static and dynamic processing pathways is optimal for both transfer tasks, and provide analyses indicating how the pathways differ. Taken in context, our results suggest that deep neural embeddings are a promising approach to unsupervised visual learning across a wide variety of domains.Preprint. Under review.