Behavior-constrained support vector machines for fMRI data analysis

Modulation Effect of Acupuncture on Functional Brain Networks and Classification of Its Manipulation With EEG Signals

Acupuncture manipulation is the key of Chinese medicine acupuncture therapy. In clinical practice, different acupuncture manipulations are required to achieve different therapeutic effects, which means it is crucial to distinguish different acupuncture manipulations. In this paper, we proposed a classification framework for different acupuncture manipulations, which employed the graph theory and machine learning method. Multichannel EEG signals evoked by acupuncture at "Zusanli" acupoint were recorded from healthy humans by two acupuncture manipulations: twirling-rotating (TR) and lifting-thrusting (LT). Phase locking value was used to estimate the phase synchronization of pair-wise EEG channels. It was found that acupunctured by TR manipulation exhibit significantly higher synchronization degree than acupunctured by LT manipulation. With the construction of functional brain network, the topological features of graph theory were extracted. Taken the network features as inputs, machine learning classifiers were established to classify acupuncture manipulations. The highest accuracy can achieve 92.14% with support vector machine. By further optimizing the network features utilized in machine learning classifiers, it was found that the combination of node betweenness and small world network index is the most effective factor for acupuncture manipulations classification. These findings suggested that our approach provides new ideas for automatically identify acupuncture manipulations from the perspective of functional brain networks and machine learning methods.

Using human brain activity to guide machine learning

Machine learning is a field of computer science that builds algorithms that learn. In many cases, machine learning algorithms are used to recreate a human ability like adding a caption to a photo, driving a car, or playing a game. While the human brain has long served as a source of inspiration for machine learning, little effort has been made to directly use data collected from working brains as a guide for machine learning algorithms. Here we demonstrate a new paradigm of “neurally-weighted” machine learning, which takes fMRI measurements of human brain activity from subjects viewing images, and infuses these data into the training process of an object recognition learning algorithm to make it more consistent with the human brain. After training, these neurally-weighted classifiers are able to classify images without requiring any additional neural data. We show that our neural-weighting approach can lead to large performance gains when used with traditional machine vision features, as well as to significant improvements with already high-performing convolutional neural network features. The effectiveness of this approach points to a path forward for a new class of hybrid machine learning algorithms which take both inspiration and direct constraints from neuronal data.

A Hybrid LDA+gCCA Model for fMRI Data Classification and Visualization

Linear predictive models are applied to functional MRI (fMRI) data to estimate boundaries that predict experimental task states for scans. These boundaries are visualized as statistical parametric maps (SPMs) and range from low to high spatial reproducibility across subjects (e.g., Strother , 2004; LaConte , 2003). Such inter-subject pattern reproducibility is an essential characteristic of interpretable SPMs that generalize across subjects. Therefore, we introduce a flexible hybrid model that optimizes reproducibility by simultaneously enhancing the prediction power and reproducibility. This hybrid model is formed by a weighted summation of the optimization functions of a linear discriminate analysis (LDA) model and a generalized canonical correlation (gCCA) model (Afshin-Pour , 2012). LDA preserves the model's ability to discriminate the fMRI scans of multiple brain states while gCCA finds a linear combination for each subject's scans such that the estimated boundary map is reproducible. The hybrid model is implemented in a split-half resampling framework (Strother , 2010) which provides reproducibility (r) and prediction (p) quality metrics. Then the model was compared with LDA, and Gaussian Naive Bayes (GNB). For simulated fMRI data, the hybrid model outperforms the other two techniques in terms of receiver operating characteristic (ROC) curves, particularly for detecting less predictable but spatially reproducible networks. These techniques were applied to real fMRI data to estimate the maps for two task contrasts. Our results indicate that compared to LDA and GNB, the hybrid model can provide maps with large increases in reproducibility for small reductions in prediction, which are jointly closer to the ideal performance point of (p=1, r=1).

Perceptual Annotation: Measuring Human Vision to Improve Computer Vision

For many problems in computer vision, human learners are considerably better than machines. Humans possess highly accurate internal recognition and learning mechanisms that are not yet understood, and they frequently have access to more extensive training data through a lifetime of unbiased experience with the visual world. We propose to use visual psychophysics to directly leverage the abilities of human subjects to build better machine learning systems. First, we use an advanced online psychometric testing platform to make new kinds of annotation data available for learning. Second, we develop a technique for harnessing these new kinds of information-"perceptual annotations"-for support vector machines. A key intuition for this approach is that while it may remain infeasible to dramatically increase the amount of data and high-quality labels available for the training of a given system, measuring the exemplar-by-exemplar difficulty and pattern of errors of human annotators can provide important information for regularizing the solution of the system at hand. A case study for the problem face detection demonstrates that this approach yields state-of-the-art results on the challenging FDDB data set.

Shifted encoding strategy in retinal luminance adaptation: from firing rate to neural correlation

Neuronal responses to prolonged stimulation attenuate over time. Here, we ask a fundamental question: is adaptation a simple process for the neural system during which sustained input is ignored, or is it actually part of a strategy for the neural system to adjust its encoding properties dynamically? After simultaneously recording the activities of a group of bullfrog's retinal ganglion cells (dimming detectors) in response to sustained dimming stimulation, we applied a combination of information analysis approaches to explore the time-dependent nature of information encoding during the adaptation. We found that at the early stage of the adaptation, the stimulus information was mainly encoded in firing rates, whereas at the late stage of the adaptation, it was more encoded in neural correlations. Such a transition in encoding properties is not a simple consequence of the attenuation of neuronal firing rates, but rather involves an active change in the neural correlation strengths, suggesting that it is a strategy adopted by the neural system for functional purposes. Our results reveal that in encoding a prolonged stimulation, the neural system may utilize concerted, but less active, firings of neurons to encode information.

Functional magnetic resonance imaging for imaging neural activity in the human brain: the annual progress

Functional magnetic resonance imaging (fMRI) is recently developed and applied to measure the hemodynamic response related to neural activity. The fMRI can not only noninvasively record brain signals without risks of ionising radiation inherent in other scanning methods, such as CT or PET scans, but also record signal from all regions of the brain, unlike EEG/MEG which are biased towards the cortical surface. This paper introduces the fundamental principles and summarizes the research progress of the last year for imaging neural activity in the human brain. Aims of functional analysis of neural activity from fMRI include biological findings, functional connectivity, vision and hearing research, emotional research, neurosurgical planning, pain management, and many others. Besides formulations and basic processing methods, models and strategies of processing technology are introduced, including general linear model, nonlinear model, generative model, spatial pattern analysis, statistical analysis, correlation analysis, and multimodal combination. This paper provides readers the most recent representative contributions in the area.