Stimulus-related independent component and voxel-wise analysis of human brain activity during free viewing of a feature film

Understanding how the brain processes stimuli in a rich natural environment is a fundamental goal of neuroscience. Here, we showed a feature film to 10 healthy volunteers during functional magnetic resonance imaging (fMRI) of hemodynamic brain activity. We then annotated auditory and visual features of the motion picture to inform analysis of the hemodynamic data. The annotations were fitted to both voxel-wise data and brain network time courses extracted by independent component analysis (ICA). Auditory annotations correlated with two independent components (IC) disclosing two functional networks, one responding to variety of auditory stimulation and another responding preferentially to speech but parts of the network also responding to non-verbal communication. Visual feature annotations correlated with four ICs delineating visual areas according to their sensitivity to different visual stimulus features. In comparison, a separate voxel-wise general linear model based analysis disclosed brain areas preferentially responding to sound energy, speech, music, visual contrast edges, body motion and hand motion which largely overlapped the results revealed by ICA. Differences between the results of IC- and voxel-based analyses demonstrate that thorough analysis of voxel time courses is important for understanding the activity of specific sub-areas of the functional networks, while ICA is a valuable tool for revealing novel information about functional connectivity which need not be explained by the predefined model. Our results encourage the use of naturalistic stimuli and tasks in cognitive neuroimaging to study how the brain processes stimuli in rich natural environments.     

Invariant recognition of feature combinations in the visual system

 The operation of a hierarchical competitive network model (VisNet) of invariance learning in the visual system is investigated to determine how this class of architecture can solve problems that require the spatial binding of features. First, we show that VisNet neurons can be trained to provide transform-invariant discriminative responses to stimuli which are composed of the same basic alphabet of features, where no single stimulus contains a unique feature not shared by any other stimulus. The investigation shows that the network can discriminate stimuli consisting of sets of features which are subsets or supersets of each other. Second, a key feature-binding issue we address is how invariant representations of low-order combinations of features in the early layers of the visual system are able to uniquely specify the correct spatial arrangement of features in the overall stimulus and ensure correct stimulus identification in the output layer. We show that output layer neurons can learn new stimuli if the lower layers are trained solely through exposure to simpler feature combinations from which the new stimuli are composed. Moreover, we show that after training on the low-order feature combinations which are common to many objects, this architecture can – after training with a whole stimulus in some locations – generalise correctly to the same stimulus when it is shown in a new location. We conclude that this type of hierarchical model can solve feature-binding problems to produce correct invariant identification of whole stimuli. Received: 4 August 1999 / Accepted in revised form: 11 October 2000     

FrankSum: new feature selection method for protein function prediction

In the study of in silico functional genomics, improving the performance of protein function prediction is the ultimate goal for identifying proteins associated with defined cellular functions. The classical prediction approach is to employ pairwise sequence alignments. However this method often faces difficulties when no statistically significant homologous sequences are identified. An alternative way is to predict protein function from sequence-derived features using machine learning. In this case the choice of possible features which can be derived from the sequence is of vital importance to ensure adequate discrimination to predict function. In this paper we have successfully selected biologically significant features for protein function prediction. This was performed using a new feature selection method (FrankSum) that avoids data distribution assumptions, uses a data independent measurement (p-value) within the feature, identifies redundancy between features and uses an appropriate ranking criterion for feature selection. We have shown that classifiers generated from features selected by FrankSum outperforms classifiers generated from full feature sets, randomly selected features and features selected from the Wrapper method. We have also shown the features are concordant across all species and top ranking features are biologically informative. We conclude that feature selection is vital for successful protein function prediction and FrankSum is one of the feature selection methods that can be applied successfully to such a domain.     

Impact of load-related neural processes on feature binding in visuospatial working memory

Background

The capacity of visual working memory (WM) is substantially limited and only a fraction of what we see is maintained as a temporary trace. The process of binding visual features has been proposed as an adaptive means of minimising information demands on WM. However the neural mechanisms underlying this process, and its modulation by task and load effects, are not well understood.

Objective

To investigate the neural correlates of feature binding and its modulation by WM load during the sequential phases of encoding, maintenance and retrieval.

Methods and Findings

18 young healthy participants performed a visuospatial WM task with independent factors of load and feature conjunction (object identity and position) in an event-related functional MRI study. During stimulus encoding, load-invariant conjunction-related activity was observed in left prefrontal cortex and left hippocampus. During maintenance, greater activity for task demands of feature conjunction versus single features, and for increased load was observed in left-sided regions of the superior occipital cortex, precuneus and superior frontal cortex. Where these effects were expressed in overlapping cortical regions, their combined effect was additive. During retrieval, however, an interaction of load and feature conjunction was observed. This modulation of feature conjunction activity under increased load was expressed through greater deactivation in medial structures identified as part of the default mode network.

Conclusions and Significance

The relationship between memory load and feature binding qualitatively differed through each phase of the WM task. Of particular interest was the interaction of these factors observed within regions of the default mode network during retrieval which we interpret as suggesting that at low loads, binding processes may be ‘automatic’ but at higher loads it becomes a resource-intensive process leading to disengagement of activity in this network. These findings provide new insights into how feature binding operates within the capacity-limited WM system.     

Binding by temporal structure in multiple feature domains of an oscillatory neuronal network

An important step in visual processing is the segregation of objects in a visual scene from one another and from the embedding background. According to current theories of visual neuroscience, the different features of a particular object are represented by cells which are spatially distributed across multiple visual areas in the brain. The segregation of an object therefore requires the unique identification and integration of the pertaining cells which have to be “bound” into one assembly coding for the object in question. Several authors have suggested that such a binding of cells could be achieved by the selective synchronization of temporally structured responses of the neurons activated by features of the same stimulus. This concept has recently gained support by the observation of stimulus-dependent oscillatory activity in the visual system of the cat, pigeon and monkey. Furthermore, experimental evidence has been found for the formation and segregation of synchronously active cell assemblies representing different stimuli in the visual field. In this study, we investigate temporally structured activity in networks with single and multiple feature domains. As a first step, we examine the formation and segregation of cell assemblies by synchronizing and desynchronizing connections within a single feature module. We then demonstrate that distributed assemblies can be appropriately bound in a network comprising three modules selective for stimulus disparity, orientation and colour, respectively. In this context, we address the principal problem of segregating assemblies representing spatially overlapping stimuli in a distributed architecture. Using synchronizing as well as desynchronizing mechanisms, our simulations demonstrate that the binding problem can be solved by temporally correlated responses of cells which are distributed across multiple feature modules. Received: 25 March 1993/Accepted in revised form: 8 September 1993     

Coevolution of active vision and feature selection

We show that complex visual tasks, such as position- and size-invariant shape recognition and navigation in the environment, can be tackled with simple architectures generated by a coevolutionary process of active vision and feature selection. Behavioral machines equipped with primitive vision systems and direct pathways between visual and motor neurons are evolved while they freely interact with their environments. We describe the application of this methodology in three sets of experiments, namely, shape discrimination, car driving, and robot navigation. We show that these systems develop sensitivity to a number of oriented, retinotopic, visual-feature-oriented edges, corners, height, and a behavioral repertoire to locate, bring, and keep these features in sensitive regions of the vision system, resembling strategies observed in simple insects.     

Quality control metrics for LC-MS feature detection tools demonstrated on Saccharomyces cerevisiae proteomic profiles

Quantitative proteomic profiling using liquid chromatography-mass spectrometry is emerging as an important tool for biomarker discovery, prompting development of algorithms for high-throughput peptide feature detection in complex samples. However, neither annotated standard data sets nor quality control metrics currently exist for assessing the validity of feature detection algorithms. We propose a quality control metric, Mass Deviance, for assessing the accuracy of feature detection tools. Because the Mass Deviance metric is derived from the natural distribution of peptide masses, it is machine- and proteome-independent and enables assessment of feature detection tools in the absence of completely annotated data sets. We validate the use of Mass Deviance with a second, independent metric that is based on isotopic distributions, demonstrating that we can use Mass Deviance to identify aberrant features with high accuracy. We then demonstrate the use of independent metrics in tandem as a robust way to evaluate the performance of peptide feature detection algorithms. This work is done on complex LC-MS profiles of Saccharomyces cerevisiae which present a significant challenge to peptide feature detection algorithms.     

CBFS: high performance feature selection algorithm based on feature clearness

Background

The goal of feature selection is to select useful features and simultaneously exclude garbage features from a given dataset for classification purposes. This is expected to bring reduction of processing time and improvement of classification accuracy.

Methodology

In this study, we devised a new feature selection algorithm (CBFS) based on clearness of features. Feature clearness expresses separability among classes in a feature. Highly clear features contribute towards obtaining high classification accuracy. CScore is a measure to score clearness of each feature and is based on clustered samples to centroid of classes in a feature. We also suggest combining CBFS and other algorithms to improve classification accuracy.

Conclusions/Significance

From the experiment we confirm that CBFS is more excellent than up-to-date feature selection algorithms including FeaLect. CBFS can be applied to microarray gene selection, text categorization, and image classification.     

A computational model for feature binding

The "Binding Problem" is an important problem across many disciplines, including psychology, neuroscience, computational modeling, and even philosophy. In this work, we proposed a novel computational model, Bayesian Linking Field Model, for feature binding in visual perception, by combining the idea of noisy neuron model, Bayesian method, Linking Field Network and competitive mechanism. Simulation Experiments demonstrated that our model perfectly fulfilled the task of feature binding in visual perception and provided us some enlightening idea for future research.     

A stable biologically motivated learning mechanism for visual feature extraction to handle facial categorization

The brain mechanism of extracting visual features for recognizing various objects has consistently been a controversial issue in computational models of object recognition. To extract visual features, we introduce a new, biologically motivated model for facial categorization, which is an extension of the Hubel and Wiesel simple-to-complex cell hierarchy. To address the synaptic stability versus plasticity dilemma, we apply the Adaptive Resonance Theory (ART) for extracting informative intermediate level visual features during the learning process, which also makes this model stable against the destruction of previously learned information while learning new information. Such a mechanism has been suggested to be embedded within known laminar microcircuits of the cerebral cortex. To reveal the strength of the proposed visual feature learning mechanism, we show that when we use this mechanism in the training process of a well-known biologically motivated object recognition model (the HMAX model), it performs better than the HMAX model in face/non-face classification tasks. Furthermore, we demonstrate that our proposed mechanism is capable of following similar trends in performance as humans in a psychophysical experiment using a face versus non-face rapid categorization task.     

Novel unsupervised feature filtering of biological data

MOTIVATION: Many methods have been developed for selecting small informative feature subsets in large noisy data. However, unsupervised methods are scarce. Examples are using the variance of data collected for each feature, or the projection of the feature on the first principal component. We propose a novel unsupervised criterion, based on SVD-entropy, selecting a feature according to its contribution to the entropy (CE) calculated on a leave-one-out basis. This can be implemented in four ways: simple ranking according to CE values (SR); forward selection by accumulating features according to which set produces highest entropy (FS1); forward selection by accumulating features through the choice of the best CE out of the remaining ones (FS2); backward elimination (BE) of features with the lowest CE. RESULTS: We apply our methods to different benchmarks. In each case we evaluate the success of clustering the data in the selected feature spaces, by measuring Jaccard scores with respect to known classifications. We demonstrate that feature filtering according to CE outperforms the variance method and gene-shaving. There are cases where the analysis, based on a small set of selected features, outperforms the best score reported when all information was used. Our method calls for an optimal size of the relevant feature set. This turns out to be just a few percents of the number of genes in the two Leukemia datasets that we have analyzed. Moreover, the most favored selected genes turn out to have significant GO enrichment in relevant cellular processes.     

Protein sequence classification using feature hashing

Recent advances in next-generation sequencing technologies have resulted in an exponential increase in the rate at which protein sequence data are being acquired. The k-gram feature representation, commonly used for protein sequence classification, usually results in prohibitively high dimensional input spaces, for large values of k. Applying data mining algorithms to these input spaces may be intractable due to the large number of dimensions. Hence, using dimensionality reduction techniques can be crucial for the performance and the complexity of the learning algorithms. In this paper, we study the applicability of feature hashing to protein sequence classification, where the original high-dimensional space is "reduced" by hashing the features into a low-dimensional space, using a hash function, i.e., by mapping features into hash keys, where multiple features can be mapped (at random) to the same hash key, and "aggregating" their counts. We compare feature hashing with the "bag of k-grams" approach. Our results show that feature hashing is an effective approach to reducing dimensionality on protein sequence classification tasks.     

Adaptive pattern classification and universal recoding: I. Parallel development and coding of neural feature detectors

This paper analyses a model for the parallel development and adult coding of neural feature detectors. The model was introduced in Grossberg (1976). We show how experience can retune feature detectors to respond to a prescribed convex set of spatial patterns. In particular, the detectors automatically respond to average features chosen from the set even if the average features have never been experienced. Using this procedure, any set of arbitrary spatial patterns can be recoded, or transformed, into any other spatial patterns (universal recoding), if there are sufficiently many cells in the network's cortex. The network is built from short term memory (STM) and long term memory (LTM) mechanisms, including mechanisms of adaptation, filtering, contrast enhancement, tuning, and nonspecific arousal. These mechanisms capture some experimental properties of plasticity in the kitten visual cortex. The model also suggests a classification of adult feature detector properties in terms of a small number of functional principles. In particular, experiments on retinal dynamics, including amarcrine cell function, are suggested.Supported in part by the Advanced Research Projects Agency under ONR Contract No. N00014-76-C-0185     

Vessel feature changes as a tool for detecting annual rings in tropical trees

Radial variations in vessel features (vessel lumen area and frequency) of four tropical tree species grown in tropical savanna, monsoon, and rainforest climates were investigated to detect indistinct annual rings. Leaf and soil water potentials were measured periodically to show annual variations in water availability and their relationship to radial variations in vessel features. In addition, the accuracy of annual-ring detection was estimated using trees of known age. Radial variations in vessel features showed annual cyclicity in all three sites. The vessel feature that showed clear annual cyclicity was different among the species and sites. Furthermore, the variation pattern of vessel features in the rainforest site tended to be different between two radii within individual trees, implying low synchronicity of wood formation in the wet area. Compared with leaf and soil water potentials, vessel features varied independent of water availability in all species in the rainforest site and most species in the savanna and monsoon sites. The direct effect of low water availability on vessel features was considered only in one species in the savanna and monsoon sites. Nevertheless, the deviation of vessel feature cycles from tree age was similar and reasonably small (mostly within ±10 %) among all sites. These results indicated a circannual rhythm in vessel formation. Thus, vessel feature changes seemed to aid in detecting annual rings in trees throughout wide tropical areas; however, asynchronous wood formation must be considered in dendrochronology, especially in humid tropics.     

Size matters: large objects capture attention in visual search

Can objects or events ever capture one''s attention in a purely stimulus-driven manner? A recent review of the literature set out the criteria required to find stimulus-driven attentional capture independent of goal-directed influences, and concluded that no published study has satisfied that criteria. Here visual search experiments assessed whether an irrelevantly large object can capture attention. Capture of attention by this static visual feature was found. The results suggest that a large object can indeed capture attention in a stimulus-driven manner and independent of displaywide features of the task that might encourage a goal-directed bias for large items. It is concluded that these results are either consistent with the stimulus-driven criteria published previously or alternatively consistent with a flexible, goal-directed mechanism of saliency detection.     

The theory of feature systems: One feature versus two for Kayardild tense-aspect-mood

Features are central to all major theories of syntax and morphology. Yet it can be a non-trivial task to determine the inventory of features and their values for a given language, and in particular to determine whether to postulate one feature or two in the same semantico-syntactic domain. We illustrate this from tense-aspect-mood (TAM) in Kayardild, and adduce principles for deciding in general between one-feature and two-feature analyses, thereby contributing to the theory of feature systems and their typology.Kayardild shows striking inflectional complexities, investigated in two major studies (Evans 1995a; Round 2013), and it proves particularly revealing for our topic. Both Evans and Round claimed that clauses in Kayardild have not one but two concurrent TAM features. While it is perfectly possible for a language to have two features of the same type, it is unusual. Accordingly, we establish general arguments which would justify postulating two features rather than one; we then apply these specifically to Kayardild TAM. Our finding is at variance with both Evans and Round; on all counts, the evidence which would motivate an analysis in terms of one TAM feature or two is either approximately balanced, or clearly favours an analysis with just one.Thus even when faced with highly complex language facts, we can apply a principled approach to the question of whether we are dealing with one feature or two, and this is encouraging for the many of us seeking a rigorous science of typology. We also find that Kayardild, which in many ways is excitingly exotic, is in this one corner of its grammar quite ordinary.     

Feature-Specific Encoding Flexibility in Visual Working Memory

The current study examined selective encoding in visual working memory by systematically investigating interference from task-irrelevant features. The stimuli were objects defined by three features (color, shape, and location), and during a delay period, any of the features could switch between two objects. Additionally, single- and whole-probe trials were randomized within experimental blocks to investigate effects of memory retrieval. A series of relevant-feature switch detection tasks, where one feature was task-irrelevant, showed that interference from the task-irrelevant feature was only observed in the color-shape task, suggesting that color and shape information could be successfully filtered out, but location information could not, even when location was a task-irrelevant feature. Therefore, although location information is added to object representations independent of task demands in a relatively automatic manner, other features (e.g., color, shape) can be flexibly added to object representations.