Function approximation in inhibitory networks

Modeling the Role of Contour Integration in Visual Inference

Under difficult viewing conditions, the brain's visual system uses a variety of recurrent modulatory mechanisms to augment feedforward processing. One resulting phenomenon is contour integration, which occurs in the primary visual (V1) cortex and strengthens neural responses to edges if they belong to a larger smooth contour. Computational models have contributed to an understanding of the circuit mechanisms of contour integration, but less is known about its role in visual perception. To address this gap, we embedded a biologically grounded model of contour integration in a task-driven artificial neural network and trained it using a gradient-descent variant. We used this model to explore how brain-like contour integration may be optimized for high-level visual objectives as well as its potential roles in perception. When the model was trained to detect contours in a background of random edges, a task commonly used to examine contour integration in the brain, it closely mirrored the brain in terms of behavior, neural responses, and lateral connection patterns. When trained on natural images, the model enhanced weaker contours and distinguished whether two points lay on the same versus different contours. The model learned robust features that generalized well to out-of-training-distribution stimuli. Surprisingly, and in contrast with the synthetic task, a parameter-matched control network without recurrence performed the same as or better than the model on the natural-image tasks. Thus, a contour integration mechanism is not essential to perform these more naturalistic contour-related tasks. Finally, the best performance in all tasks was achieved by a modified contour integration model that did not distinguish between excitatory and inhibitory neurons.

Distinctive properties of biological neural networks and recent advances in bottom-up approaches toward a better biologically plausible neural network

Although it may appear infeasible and impractical, building artificial intelligence (AI) using a bottom-up approach based on the understanding of neuroscience is straightforward. The lack of a generalized governing principle for biological neural networks (BNNs) forces us to address this problem by converting piecemeal information on the diverse features of neurons, synapses, and neural circuits into AI. In this review, we described recent attempts to build a biologically plausible neural network by following neuroscientifically similar strategies of neural network optimization or by implanting the outcome of the optimization, such as the properties of single computational units and the characteristics of the network architecture. In addition, we proposed a formalism of the relationship between the set of objectives that neural networks attempt to achieve, and neural network classes categorized by how closely their architectural features resemble those of BNN. This formalism is expected to define the potential roles of top-down and bottom-up approaches for building a biologically plausible neural network and offer a map helping the navigation of the gap between neuroscience and AI engineering.

The plasticitome of cortical interneurons

Hebb postulated that, to store information in the brain, assemblies of excitatory neurons coding for a percept are bound together via associative long-term synaptic plasticity. In this view, it is unclear what role, if any, is carried out by inhibitory interneurons. Indeed, some have argued that inhibitory interneurons are not plastic. Yet numerous recent studies have demonstrated that, similar to excitatory neurons, inhibitory interneurons also undergo long-term plasticity. Here, we discuss the many diverse forms of long-term plasticity that are found at inputs to and outputs from several types of cortical inhibitory interneuron, including their plasticity of intrinsic excitability and their homeostatic plasticity. We explain key plasticity terminology, highlight key interneuron plasticity mechanisms, extract overarching principles and point out implications for healthy brain functionality as well as for neuropathology. We introduce the concept of the plasticitome - the synaptic plasticity counterpart to the genome or the connectome - as well as nomenclature and definitions for dealing with this rich diversity of plasticity. We argue that the great diversity of interneuron plasticity rules is best understood at the circuit level, for example as a way of elucidating how the credit-assignment problem is solved in deep biological neural networks.

Buice M. MouseNet: A biologically constrained convolutional neural network model for the mouse visual cortex

Convolutional neural networks trained on object recognition derive inspiration from the neural architecture of the visual system in mammals, and have been used as models of the feedforward computation performed in the primate ventral stream. In contrast to the deep hierarchical organization of primates, the visual system of the mouse has a shallower arrangement. Since mice and primates are both capable of visually guided behavior, this raises questions about the role of architecture in neural computation. In this work, we introduce a novel framework for building a biologically constrained convolutional neural network model of the mouse visual cortex. The architecture and structural parameters of the network are derived from experimental measurements, specifically the 100-micrometer resolution interareal connectome, the estimates of numbers of neurons in each area and cortical layer, and the statistics of connections between cortical layers. This network is constructed to support detailed task-optimized models of mouse visual cortex, with neural populations that can be compared to specific corresponding populations in the mouse brain. Using a well-studied image classification task as our working example, we demonstrate the computational capability of this mouse-sized network. Given its relatively small size, MouseNet achieves roughly 2/3rds the performance level on ImageNet as VGG16. In combination with the large scale Allen Brain Observatory Visual Coding dataset, we use representational similarity analysis to quantify the extent to which MouseNet recapitulates the neural representation in mouse visual cortex. Importantly, we provide evidence that optimizing for task performance does not improve similarity to the corresponding biological system beyond a certain point. We demonstrate that the distributions of some physiological quantities are closer to the observed distributions in the mouse brain after task training. We encourage the use of the MouseNet architecture by making the code freely available.

A Feed-Forward Neural Network for Increasing the Hopfield-Network Storage Capacity

In the hippocampal dentate gyrus (DG), pattern separation mainly depends on the concepts of 'expansion recoding', meaning random mixing of different DG input channels. However, recent advances in neurophysiology have challenged the theory of pattern separation based on these concepts. In this study, we propose a novel feed-forward neural network, inspired by the structure of the DG and neural oscillatory analysis, to increase the Hopfield-network storage capacity. Unlike the previously published feed-forward neural networks, our bio-inspired neural network is designed to take advantage of both biological structure and functions of the DG. To better understand the computational principles of pattern separation in the DG, we have established a mouse model of environmental enrichment. We obtained a possible computational model of the DG, associated with better pattern separation ability, by using neural oscillatory analysis. Furthermore, we have developed a new algorithm based on Hebbian learning and coupling direction of neural oscillation to train the proposed neural network. The simulation results show that our proposed network significantly expands the storage capacity of Hopfield network, and more effective pattern separation is achieved. The storage capacity rises from 0.13 for the standard Hopfield network to 0.32 using our model when the overlap in patterns is 10%.

Engineering a Less Artificial Intelligence

Despite enormous progress in machine learning, artificial neural networks still lag behind brains in their ability to generalize to new situations. Given identical training data, differences in generalization are caused by many defining features of a learning algorithm, such as network architecture and learning rule. Their joint effect, called "inductive bias," determines how well any learning algorithm-or brain-generalizes: robust generalization needs good inductive biases. Artificial networks use rather nonspecific biases and often latch onto patterns that are only informative about the statistics of the training data but may not generalize to different scenarios. Brains, on the other hand, generalize across comparatively drastic changes in the sensory input all the time. We highlight some shortcomings of state-of-the-art learning algorithms compared to biological brains and discuss several ideas about how neuroscience can guide the quest for better inductive biases by providing useful constraints on representations and network architecture.

Approximating the Architecture of Visual Cortex in a Convolutional Network

Deep convolutional neural networks (CNNs) have certain structural, mechanistic, representational, and functional parallels with primate visual cortex and also many differences. However, perhaps some of the differences can be reconciled. This study develops a cortex-like CNN architecture, via (1) a loss function that quantifies the consistency of a CNN architecture with neural data from tract tracing, cell reconstruction, and electrophysiology studies; (2) a hyperparameter-optimization approach for reducing this loss, and (3) heuristics for organizing units into convolutional-layer grids. The optimized hyperparameters are consistent with neural data. The cortex-like architecture differs from typical CNN architectures. In particular, it has longer skip connections, larger kernels and strides, and qualitatively different connection sparsity. Importantly, layers of the cortex-like network have one-to-one correspondences with cortical neuron populations. This should allow unambiguous comparison of model and brain representations in the future and, consequently, more precise measurement of progress toward more biologically realistic deep networks.

Strategies to develop robust neural network models: Prediction of flash point as a case study

Artificial neural network (ANN) is one of the most widely used methods to develop accurate predictive models based on artificial intelligence and machine learning. In the present study, the important practical aspects of developing a reliable ANN model e.g. appropriate assignment of the number of neurons, number of hidden layers, transfer function, training algorithm, dataset division and initialization of the network are discussed. As a case study, predictability of the flash point for a dataset of 740 organic compounds using ANNs was investigated via a total number of 484220ANNs to allow covering a wide range of parameters affecting the performance of an ANN. Among all studied parameters, the number of neurons or layers was found to be the most important parameters to develop a reliable ANN with low overfitting risk. To evaluate appropriate number of neurons and layers, a value of equal or greater than 10 for the ratio of the training samples to the ANN constants was suggested as a rule of thumb. More ever, a strategy for evaluation of the authentic performance of ANNs and deciding about the reliability of an ANN model was proposed which is applicable to other models developed by supervised learning. Based on the introduced considerations, an ANN model was proposed for predicting the flash point of pure organic compounds. According to the results, the new model was found to produce the lowest error compared to other available models.

Toward an Integration of Deep Learning and Neuroscience

Neuroscience has focused on the detailed implementation of computation, studying neural codes, dynamics and circuits. In machine learning, however, artificial neural networks tend to eschew precisely designed codes, dynamics or circuits in favor of brute force optimization of a cost function, often using simple and relatively uniform initial architectures. Two recent developments have emerged within machine learning that create an opportunity to connect these seemingly divergent perspectives. First, structured architectures are used, including dedicated systems for attention, recursion and various forms of short- and long-term memory storage. Second, cost functions and training procedures have become more complex and are varied across layers and over time. Here we think about the brain in terms of these ideas. We hypothesize that (1) the brain optimizes cost functions, (2) the cost functions are diverse and differ across brain locations and over development, and (3) optimization operates within a pre-structured architecture matched to the computational problems posed by behavior. In support of these hypotheses, we argue that a range of implementations of credit assignment through multiple layers of neurons are compatible with our current knowledge of neural circuitry, and that the brain's specialized systems can be interpreted as enabling efficient optimization for specific problem classes. Such a heterogeneously optimized system, enabled by a series of interacting cost functions, serves to make learning data-efficient and precisely targeted to the needs of the organism. We suggest directions by which neuroscience could seek to refine and test these hypotheses.