Learning to represent visual input

Behavioral Studies Using Large-Scale Brain Networks – Methods and Validations

Mapping human behaviors to brain activity has become a key focus in modern cognitive neuroscience. As methods such as functional MRI (fMRI) advance cognitive scientists show an increasing interest in investigating neural activity in terms of functional connectivity and brain networks, rather than activation in a single brain region. Due to the noisy nature of neural activity, determining how behaviors are associated with specific neural signals is not well-established. Previous research has suggested graph theory techniques as a solution. Graph theory provides an opportunity to interpret human behaviors in terms of the topological organization of brain network architecture. Graph theory-based approaches, however, only scratch the surface of what neural connections relate to human behavior. Recently, the development of data-driven methods, e.g., machine learning and deep learning approaches, provide a new perspective to study the relationship between brain networks and human behaviors across the whole brain, expanding upon past literatures. In this review, we sought to revisit these data-driven approaches to facilitate our understanding of neural mechanisms and build models of human behaviors. We start with the popular graph theory approach and then discuss other data-driven approaches such as connectome-based predictive modeling, multivariate pattern analysis, network dynamic modeling, and deep learning techniques that quantify meaningful networks and connectivity related to cognition and behaviors. Importantly, for each topic, we discuss the pros and cons of the methods in addition to providing examples using our own data for each technique to describe how these methods can be applied to real-world neuroimaging data.

Single-Input Multi-Output U-Net for Automated 2D Foetal Brain Segmentation of MR Images

In this work, we develop the Single-Input Multi-Output U-Net (SIMOU-Net), a hybrid network for foetal brain segmentation inspired by the original U-Net fused with the holistically nested edge detection (HED) network. The SIMOU-Net is similar to the original U-Net but it has a deeper architecture and takes account of the features extracted from each side output. It acts similar to an ensemble neural network, however, instead of averaging the outputs from several independently trained models, which is computationally expensive, our approach combines outputs from a single network to reduce the variance of predications and generalization errors. Experimental results using 200 normal foetal brains consisting of over 11,500 2D images produced Dice and Jaccard coefficients of 94.2 ± 5.9% and 88.7 ± 6.9%, respectively. We further tested the proposed network on 54 abnormal cases (over 3500 images) and achieved Dice and Jaccard coefficients of 91.2 ± 6.8% and 85.7 ± 6.6%, respectively.

Processing of auditory novelty across the cortical hierarchy: An intracranial electrophysiology study

Under the predictive coding hypothesis, specific spatiotemporal patterns of cortical activation are postulated to occur during sensory processing as expectations generate feedback predictions and prediction errors generate feedforward signals. Establishing experimental evidence for this information flow within cortical hierarchy has been difficult, especially in humans, due to spatial and temporal limitations of non-invasive measures of cortical activity. This study investigated cortical responses to auditory novelty using the local/global deviant paradigm, which engages the hierarchical network underlying auditory predictive coding over short ('local deviance'; LD) and long ('global deviance'; GD) time scales. Electrocorticographic responses to auditory stimuli were obtained in neurosurgical patients from regions of interest (ROIs) including auditory, auditory-related and prefrontal cortex. LD and GD effects were assayed in averaged evoked potential (AEP) and high gamma (70-150 Hz) signals, the former likely dominated by local synaptic currents and the latter largely reflecting local spiking activity. AEP LD effects were distributed across all ROIs, with greatest percentage of significant sites in core and non-core auditory cortex. High gamma LD effects were localized primarily to auditory cortex in the superior temporal plane and on the lateral surface of the superior temporal gyrus (STG). LD effects exhibited progressively longer latencies in core, non-core, auditory-related and prefrontal cortices, consistent with feedforward signaling. The spatial distribution of AEP GD effects overlapped that of LD effects, but high gamma GD effects were more restricted to non-core areas. High gamma GD effects had shortest latencies in STG and preceded AEP GD effects in most ROIs. This latency profile, along with the paucity of high gamma GD effects in the superior temporal plane, suggest that the STG plays a prominent role in initiating novelty detection signals over long time scales. Thus, the data demonstrate distinct patterns of information flow in human cortex associated with auditory novelty detection over multiple time scales.

Change Detection in Synthetic Aperture Radar Images Based on Deep Neural Networks

This paper presents a novel change detection approach for synthetic aperture radar images based on deep learning. The approach accomplishes the detection of the changed and unchanged areas by designing a deep neural network. The main guideline is to produce a change detection map directly from two images with the trained deep neural network. The method can omit the process of generating a difference image (DI) that shows difference degrees between multitemporal synthetic aperture radar images. Thus, it can avoid the effect of the DI on the change detection results. The learning algorithm for deep architectures includes unsupervised feature learning and supervised fine-tuning to complete classification. The unsupervised feature learning aims at learning the representation of the relationships between the two images. In addition, the supervised fine-tuning aims at learning the concepts of the changed and unchanged pixels. Experiments on real data sets and theoretical analysis indicate the advantages, feasibility, and potential of the proposed method. Moreover, based on the results achieved by various traditional algorithms, respectively, deep learning can further improve the detection performance.

High-order neural networks and kernel methods for peptide-MHC binding prediction

MOTIVATION

Effective computational methods for peptide-protein binding prediction can greatly help clinical peptide vaccine search and design. However, previous computational methods fail to capture key nonlinear high-order dependencies between different amino acid positions. As a result, they often produce low-quality rankings of strong binding peptides. To solve this problem, we propose nonlinear high-order machine learning methods including high-order neural networks (HONNs) with possible deep extensions and high-order kernel support vector machines to predict major histocompatibility complex-peptide binding.

RESULTS

The proposed high-order methods improve quality of binding predictions over other prediction methods. With the proposed methods, a significant gain of up to 25-40% is observed on the benchmark and reference peptide datasets and tasks. In addition, for the first time, our experiments show that pre-training with high-order semi-restricted Boltzmann machines significantly improves the performance of feed-forward HONNs. Moreover, our experiments show that the proposed shallow HONN outperform the popular pre-trained deep neural network on most tasks, which demonstrates the effectiveness of modelling high-order feature interactions for predicting major histocompatibility complex-peptide binding.

AVAILABILITY AND IMPLEMENTATION

There is no associated distributable software.

CONTACT

renqiang@nec-labs.com or mark.gerstein@yale.edu

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Learning Separable Filters

Learning filters to produce sparse image representations in terms of over-complete dictionaries has emerged as a powerful way to create image features for many different purposes. Unfortunately, these filters are usually both numerous and non-separable, making their use computationally expensive. In this paper, we show that such filters can be computed as linear combinations of a smaller number of separable ones, thus greatly reducing the computational complexity at no cost in terms of performance. This makes filter learning approaches practical even for large images or 3D volumes, and we show that we significantly outperform state-of-the-art methods on the curvilinear structure extraction task, in terms of both accuracy and speed. Moreover, our approach is general and can be used on generic convolutional filter banks to reduce the complexity of the feature extraction step.

The interplay of plasticity and adaptation in neural circuits: a generative model

Multiple neural and synaptic phenomena take place in the brain. They operate over a broad range of timescales, and the consequences of their interplay are still unclear. In this work, I study a computational model of a recurrent neural network in which two dynamic processes take place: sensory adaptation and synaptic plasticity. Both phenomena are ubiquitous in the brain, but their dynamic interplay has not been investigated. I show that when both processes are included, the neural circuit is able to perform a specific computation: it becomes a generative model for certain distributions of input stimuli. The neural circuit is able to generate spontaneous patterns of activity that reproduce exactly the probability distribution of experienced stimuli. In particular, the landscape of the phase space includes a large number of stable states (attractors) that sample precisely this prior distribution. This work demonstrates that the interplay between distinct dynamical processes gives rise to useful computation, and proposes a framework in which neural circuit models for Bayesian inference may be developed in the future.

Pupil fluctuations track fast switching of cortical states during quiet wakefulness

Neural responses are modulated by brain state, which varies with arousal, attention, and behavior. In mice, running and whisking desynchronize the cortex and enhance sensory responses, but the quiescent periods between bouts of exploratory behaviors have not been well studied. We found that these periods of "quiet wakefulness" were characterized by state fluctuations on a timescale of 1-2 s. Small fluctuations in pupil diameter tracked these state transitions in multiple cortical areas. During dilation, the intracellular membrane potential was desynchronized, sensory responses were enhanced, and population activity was less correlated. In contrast, constriction was characterized by increased low-frequency oscillations and higher ensemble correlations. Specific subtypes of cortical interneurons were differentially activated during dilation and constriction, consistent with their participation in the observed state changes. Pupillometry has been used to index attention and mental effort in humans, but the intracellular dynamics and differences in population activity underlying this phenomenon were previously unknown.

Semi-supervised video segmentation using tree structured graphical models

We present a novel patch-based probabilistic graphical model for semi-supervised video segmentation. At the heart of our model is a temporal tree structure that links patches in adjacent frames through the video sequence. This permits exact inference of pixel labels without resorting to traditional short time window-based video processing or instantaneous decision making. The input to our algorithm is labeled key frame(s) of a video sequence and the output is pixel-wise labels along with their confidences. We propose an efficient inference scheme that performs exact inference over the temporal tree, and optionally a per frame label smoothing step using loopy BP, to estimate pixel-wise labels and their posteriors. These posteriors are used to learn pixel unaries by training a Random Decision Forest in a semi-supervised manner. These unaries are used in a second iteration of label inference to improve the segmentation quality. We demonstrate the efficacy of our proposed algorithm using several qualitative and quantitative tests on both foreground/background and multiclass video segmentation problems using publicly available and our own datasets.

Modeling language and cognition with deep unsupervised learning: a tutorial overview

Deep unsupervised learning in stochastic recurrent neural networks with many layers of hidden units is a recent breakthrough in neural computation research. These networks build a hierarchy of progressively more complex distributed representations of the sensory data by fitting a hierarchical generative model. In this article we discuss the theoretical foundations of this approach and we review key issues related to training, testing and analysis of deep networks for modeling language and cognitive processing. The classic letter and word perception problem of McClelland and Rumelhart (1981) is used as a tutorial example to illustrate how structured and abstract representations may emerge from deep generative learning. We argue that the focus on deep architectures and generative (rather than discriminative) learning represents a crucial step forward for the connectionist modeling enterprise, because it offers a more plausible model of cortical learning as well as a way to bridge the gap between emergentist connectionist models and structured Bayesian models of cognition.

Where do features come from?

It is possible to learn multiple layers of non-linear features by backpropagating error derivatives through a feedforward neural network. This is a very effective learning procedure when there is a huge amount of labeled training data, but for many learning tasks very few labeled examples are available. In an effort to overcome the need for labeled data, several different generative models were developed that learned interesting features by modeling the higher order statistical structure of a set of input vectors. One of these generative models, the restricted Boltzmann machine (RBM), has no connections between its hidden units and this makes perceptual inference and learning much simpler. More significantly, after a layer of hidden features has been learned, the activities of these features can be used as training data for another RBM. By applying this idea recursively, it is possible to learn a deep hierarchy of progressively more complicated features without requiring any labeled data. This deep hierarchy can then be treated as a feedforward neural network which can be discriminatively fine-tuned using backpropagation. Using a stack of RBMs to initialize the weights of a feedforward neural network allows backpropagation to work effectively in much deeper networks and it leads to much better generalization. A stack of RBMs can also be used to initialize a deep Boltzmann machine that has many hidden layers. Combining this initialization method with a new method for fine-tuning the weights finally leads to the first efficient way of training Boltzmann machines with many hidden layers and millions of weights.

Whatever next? Predictive brains, situated agents, and the future of cognitive science

Brains, it has recently been argued, are essentially prediction machines. They are bundles of cells that support perception and action by constantly attempting to match incoming sensory inputs with top-down expectations or predictions. This is achieved using a hierarchical generative model that aims to minimize prediction error within a bidirectional cascade of cortical processing. Such accounts offer a unifying model of perception and action, illuminate the functional role of attention, and may neatly capture the special contribution of cortical processing to adaptive success. This target article critically examines this "hierarchical prediction machine" approach, concluding that it offers the best clue yet to the shape of a unified science of mind and action. Sections 1 and 2 lay out the key elements and implications of the approach. Section 3 explores a variety of pitfalls and challenges, spanning the evidential, the methodological, and the more properly conceptual. The paper ends (sections 4 and 5) by asking how such approaches might impact our more general vision of mind, experience, and agency.

Deep Unsupervised Learning on a Desktop PC: A Primer for Cognitive Scientists

Deep belief networks hold great promise for the simulation of human cognition because they show how structured and abstract representations may emerge from probabilistic unsupervised learning. These networks build a hierarchy of progressively more complex distributed representations of the sensory data by fitting a hierarchical generative model. However, learning in deep networks typically requires big datasets and it can involve millions of connection weights, which implies that simulations on standard computers are unfeasible. Developing realistic, medium-to-large-scale learning models of cognition would therefore seem to require expertise in programing parallel-computing hardware, and this might explain why the use of this promising approach is still largely confined to the machine learning community. Here we show how simulations of deep unsupervised learning can be easily performed on a desktop PC by exploiting the processors of low cost graphic cards (graphic processor units) without any specific programing effort, thanks to the use of high-level programming routines (available in MATLAB or Python). We also show that even an entry-level graphic card can outperform a small high-performance computing cluster in terms of learning time and with no loss of learning quality. We therefore conclude that graphic card implementations pave the way for a widespread use of deep learning among cognitive scientists for modeling cognition and behavior.

Invariant Visual Object and Face Recognition: Neural and Computational Bases, and a Model, VisNet

Neurophysiological evidence for invariant representations of objects and faces in the primate inferior temporal visual cortex is described. Then a computational approach to how invariant representations are formed in the brain is described that builds on the neurophysiology. A feature hierarchy model in which invariant representations can be built by self-organizing learning based on the temporal and spatial statistics of the visual input produced by objects as they transform in the world is described. VisNet can use temporal continuity in an associative synaptic learning rule with a short-term memory trace, and/or it can use spatial continuity in continuous spatial transformation learning which does not require a temporal trace. The model of visual processing in the ventral cortical stream can build representations of objects that are invariant with respect to translation, view, size, and also lighting. The model has been extended to provide an account of invariant representations in the dorsal visual system of the global motion produced by objects such as looming, rotation, and object-based movement. The model has been extended to incorporate top-down feedback connections to model the control of attention by biased competition in, for example, spatial and object search tasks. The approach has also been extended to account for how the visual system can select single objects in complex visual scenes, and how multiple objects can be represented in a scene. The approach has also been extended to provide, with an additional layer, for the development of representations of spatial scenes of the type found in the hippocampus.

What are the Visual Features Underlying Rapid Object Recognition?

Research progress in machine vision has been very significant in recent years. Robust face detection and identification algorithms are already readily available to consumers, and modern computer vision algorithms for generic object recognition are now coping with the richness and complexity of natural visual scenes. Unlike early vision models of object recognition that emphasized the role of figure-ground segmentation and spatial information between parts, recent successful approaches are based on the computation of loose collections of image features without prior segmentation or any explicit encoding of spatial relations. While these models remain simplistic models of visual processing, they suggest that, in principle, bottom-up activation of a loose collection of image features could support the rapid recognition of natural object categories and provide an initial coarse visual representation before more complex visual routines and attentional mechanisms take place. Focusing on biologically plausible computational models of (bottom-up) pre-attentive visual recognition, we review some of the key visual features that have been described in the literature. We discuss the consistency of these feature-based representations with classical theories from visual psychology and test their ability to account for human performance on a rapid object categorization task.

Neural dynamics as sampling: a model for stochastic computation in recurrent networks of spiking neurons

The organization of computations in networks of spiking neurons in the brain is still largely unknown, in particular in view of the inherently stochastic features of their firing activity and the experimentally observed trial-to-trial variability of neural systems in the brain. In principle there exists a powerful computational framework for stochastic computations, probabilistic inference by sampling, which can explain a large number of macroscopic experimental data in neuroscience and cognitive science. But it has turned out to be surprisingly difficult to create a link between these abstract models for stochastic computations and more detailed models of the dynamics of networks of spiking neurons. Here we create such a link and show that under some conditions the stochastic firing activity of networks of spiking neurons can be interpreted as probabilistic inference via Markov chain Monte Carlo (MCMC) sampling. Since common methods for MCMC sampling in distributed systems, such as Gibbs sampling, are inconsistent with the dynamics of spiking neurons, we introduce a different approach based on non-reversible Markov chains that is able to reflect inherent temporal processes of spiking neuronal activity through a suitable choice of random variables. We propose a neural network model and show by a rigorous theoretical analysis that its neural activity implements MCMC sampling of a given distribution, both for the case of discrete and continuous time. This provides a step towards closing the gap between abstract functional models of cortical computation and more detailed models of networks of spiking neurons.

Active data collection for efficient estimation and comparison of nonlinear neural models

The stimulus-response relationship of many sensory neurons is nonlinear, but fully quantifying this relationship by a complex nonlinear model may require too much data to be experimentally tractable. Here we present a theoretical study of a general two-stage computational method that may help to significantly reduce the number of stimuli needed to obtain an accurate mathematical description of nonlinear neural responses. Our method of active data collection first adaptively generates stimuli that are optimal for estimating the parameters of competing nonlinear models and then uses these estimates to generate stimuli online that are optimal for discriminating these models. We applied our method to simple hierarchical circuit models, including nonlinear networks built on the spatiotemporal or spectral-temporal receptive fields, and confirmed that collecting data using our two-stage adaptive algorithm was far more effective for estimating and comparing competing nonlinear sensory processing models than standard nonadaptive methods using random stimuli.