Classification and recognition in artificial grammar learning: Analysis of receiver operating characteristics

In two experiments we investigated recognition and classification judgements using an artificial grammar learning paradigm. In Experiment 1, when only new test items had to be judged, analysis of z-transformed receiver operating characteristics (z-ROCs) revealed no differences between classification and recognition. In Experiment 2, where we included old test items, z-ROCs in the two tasks differed, suggesting that judgements relied on different types of information. The results are interpreted in terms of heuristics that people use when making classification and recognition judgements.     

Classification and recognition in artificial grammar learning: Analysis of receiver operating characteristics

In two experiments we investigated recognition and classification judgements using an artificial grammar learning paradigm. In Experiment 1, when only new test items had to be judged, analysis of z-transformed receiver operating characteristics (z-ROCs) revealed no differences between classification and recognition. In Experiment 2, where we included old test items, z-ROCs in the two tasks differed, suggesting that judgements relied on different types of information. The results are interpreted in terms of heuristics that people use when making classification and recognition judgements.     

Efficiency investigation from shallow to deep neural network techniques in human activity recognition

In the last years, several researchers measured different recognition rates with different artificial neural network (ANN) techniques on public data sets in the human activity recognition (HAR) problem. However an overall investigation does not exist in the literature and the efficiency of complex and deeper ANNs over shallow networks is not clear. The purpose of this paper is to investigate the recognition rate and time requirement of different kinds of ANN approaches in HAR. This work examines the performance of shallow ANN architectures with different hyper-parameters, ANN ensembles, binary ANN classifier groups, and convolutional neural networks on two public databases. Although the popularity of binary classifiers, classifier ensembles and deep learning have been significantly increasing, this study shows that shallow ANNs with appropriate hyper-parameters in combination with extracted features can reach similar or higher recognition rate in less time than other artificial neural network methods in HAR. With a well-tuned ANN we outperformed all previous results on two public databases. Consequently, instead of the more complex ANN techniques, the usage of simple ANN with two or three layers can be an appropriate choice for activity recognition.     

Neuroanatomical correlates of implicit and explicit memory for structurally possible and impossible visual objects

Implicit memory refers to nonconscious retrieval of past experience demonstrated by facilitation in test performance on tasks that do not require intentional recollection of previous experiences. Explicit memory, in contrast, refers to the conscious retrieval of prior information, as demonstrated during standard recall and recognition tasks. In this experiment, positron emission tomographic (PET) measurements of regional cerebral blood flow (CBF), a marker of local neuronal activity, were used to identify and contrast brain regions that participate in the perception, implicit memory, and explicit memory for structurally possible and impossible visual objects. Ten CBF images were acquired in 16 normal women as they made possible/impossible and old/new recognition decisions about previously studied (old) and nonstudied (new) structurally possible and impossible objects. As reported previously, object decisions for familiar possible objects were associated with increased CBF in the vicinity of the left inferior temporal and fusiform gyri and recognition memory for familiar possible objects was associated with increased CBF in the vicinity of the right hippocampus. In this report, we provide more extensive analyses of the roles of the inferior temporal cortex, the hippocampus, the parahippocampus, and the pulvinar in encoding and retrieval operations. Additionally, patterns of CBF increases and decreases provide information regarding the neural structures involved in implicit and explicit memory.     

汉词和图画再认的ERP效应及脑机制

目的：比较汉词与图画材料再认过程中的事件相关电位（ERP）表现并探讨其脑机制。方法：使用经典“学习一再认”实验模式对21名大学生进行测验,同时记录EEG,离线处理后对汉词及图画再认诱发的ERP波形主要成分及新旧效应进行比较。结果：两种材料的再认均诱发出了明显的新旧效应。顶区新旧效应头皮分布基本一致,但图画再认该效应波幅大于汉词再认;两种材料诱发的额区新旧效应有所不同,图画再认该效应为正走向,而汉词再认此效应表现为负走向趋势。结论：汉词与图画再认既存在某种共同的神经加工机制,叉分别涉及一些特异性的神经活动模式。     

The relation between recognition memory and classification learning

Two experiments investigated the relation between recognition memory and classification learning. The subjects were instructed that they would see a series of random-dot patterns and later would be asked to classify or to recognize the patterns. Following study, the subjects performed a classification task, a recognition-memory task, or both. It was found that classification-learning instructions were superior to recognition-memory instructions for the classification task, but that there was little or no effect of instructions on the recognition task. When subjects performed both recognition and classification tasks, there was no relation between saying “old” to a probe and correctly classifying it, except with old exemplars, and then only when the initial instructions had been to expect a recognition-memory test. Overall, the data show that classification and recognition can be experimentally separated. In addition, classification is often statistically independent of recognizing that items are old. These observed relations provide some constraints for the further development of models of classification learning and recognition memory.     

Understanding covert recognition

An implementation of Bruce and Young's (1986) functional model of face recognition is used to examine patterns of covert face recognition previously reported in a prosopagnosic patient, PH. Although PH is unable to recognize overly the faces of people known to him, he shows normal patterns of face processing when tested indirectly. A simple manipulation of one set of connections in the implemented model induces behaviour consistent with patterns of results from PH obtained in semantic priming and interference tasks. We compare this account with previous explanations of covert recognition and demonstrate that the implemented model provides the most natural and parsimonious account available. Two further patients are discussed who show deficits in person perception. The first (MS) is prosopagnosic but shows no covert recognition. The second (ME) is not prosopagnosic, but cannot access semantic information relating to familiar people. The model provides an account of recognition impairments which is sufficiently general also to be useful in describing these patients.     

Applications of neural networks in training science

Training science views itself as an integrated and applied science, developing practical measures founded on scientific method. Therefore, it demands consideration of a wide spectrum of approaches and methods. Especially in the field of competitive sports, research questions are usually located in complex environments, so that mainly field studies are drawn upon to obtain broad external validity. Here, the interrelations between different variables or variable sets are mostly of a nonlinear character. In these cases, methods like neural networks, e.g., the pattern recognizing methods of Self-Organizing Kohonen Feature Maps or similar instruments to identify interactions might be successfully applied to analyze data. Following on from a classification of data analysis methods in training-science research, the aim of the contribution is to give examples of varied sports in which network approaches can be effectually used in training science. First, two examples are given in which neural networks are employed for pattern recognition. While one investigation deals with the detection of sporting talent in swimming, the other is located in game sports research, identifying tactical patterns in team handball. The third and last example shows how an artificial neural network can be used to predict competitive performance in swimming.     

Decoding the Large-Scale Structure of Brain Function by Classifying Mental States Across Individuals

ABSTRACT— Brain-imaging research has largely focused on localizing patterns of activity related to specific mental processes, but recent work has shown that mental states can be identified from neuroimaging data using statistical classifiers. We investigated whether this approach could be extended to predict the mental state of an individual using a statistical classifier trained on other individuals, and whether the information gained in doing so could provide new insights into how mental processes are organized in the brain. Using a variety of classifier techniques, we achieved cross-validated classification accuracy of 80% across individuals (chance = 13%). Using a neural network classifier, we recovered a low-dimensional representation common to all the cognitive-perceptual tasks in our data set, and we used an ontology of cognitive processes to determine the cognitive concepts most related to each dimension. These results revealed a small organized set of large-scale networks that map cognitive processes across a highly diverse set of mental tasks, suggesting a novel way to characterize the neural basis of cognition.     

Functional neuroimaging studies of category specificity in object recognition: a critical review and meta-analysis

Functional neuroimaging studies in which the cortical organization for semantic knowledge has been addressed have revealed interesting dissociations in the recognition of different object categories, such as faces, natural objects, and manufactured objects. The present paper critically reviews these studies and performs a meta-analysis of stereotactic coordinates to determine whether category membership predicts patterns of brain activation across different studies. This meta-analysis revealed that, in the ventral temporal cortex, recognition of manufactured objects activates more medial aspects of the fusiform gyrus, as compared with natural object or face recognition. Face recognition activates more inferior aspects of the ventral temporal cortex, as compared with manufactured object recognition. The recognition task used—viewing, matching, or naming—also predicted brain activation patterns. Specifically, matching tasks recruit more inferior occipital regions than do either naming or viewing tasks, whereas naming tasks recruit more anterior ventral temporal sites than do either viewing or matching tasks. These findings indicate that the cognitive demands of a particular recognition task are as predictive of cortical activation patterns as is category membership.     

The variance reaction time model

The variance reaction time model (VRTM) is proposed to account for various recognition data on reaction time, the mirror effect, receiver-operating-characteristic (ROC) curves, etc. The model is based on simple and plausible assumptions within a neural network: VRTM is a two layer neural network where one layer represents items and one layer represents contexts. The recognition decision is based on a random walk of nodes activated at recognition. VRTM suggests theoretical constraints on the distributions of nodes activated at recognition and the noise in the random walk. The variability in the net inputs to nodes depends on the item frequency (the number of times that the item has been encoded) and the list length. The essential mechanism that accounts for the empirical data is a non-linear activation function. The mean activation threshold in the non-linear activation function is placed to achieve efficient discriminability between new and old items and there is variability in the activation threshold. VRTM predicts the mirror effect for low and high frequency words, a strength based mirror effect between conditions but not within one condition, appropriate ROC-curves for old/new and high/low frequency items, and list-length effects. Furthermore, it predicts appropriate means and distributions of reaction times for old/new, correct/incorrect, and high/low frequency items as well as speed/accuracy tradeoffs. VRTM has an explicit mathematical solution, it is simulated in a neural network, and it is fitted to a number of datasets.     

Artificial grammer learning by infants: an auto‐associator perspective

This paper reviews a recent article suggesting that infants use a system of algebraic rules to learn an artificial grammar (Marcus, Vijayan, Bandi Rao & Vishton, Rule learning by seven‐month‐old infants. Science, 183(1999), 77–80). In three reported experiments, infants exhibited increased responding to auditory strings that violated the pattern of elements they were habituated to. We argue that a perceptual interpretation is more parsimonious, as well as more consistent with a broad array of habituation data, and we report successful neural network simulations that implement this lower‐level interpretation. In the discussion, we discuss how our model relates to other habituation research, and how it compares to other neural network models of habituation in general, and models of the Marcus et al. (1999) task specifically.     

Effects of aging and IQ on item and associative memory

The effects of aging and IQ on performance were examined in 4 memory tasks: item recognition, associative recognition, cued recall, and free recall. For item and associative recognition, accuracy and the response time (RT) distributions for correct and error responses were explained by Ratcliff's (1978) diffusion model at the level of individual participants. The values of the components of processing identified by the model for the recognition tasks, as well as accuracy for cued and free recall, were compared across levels of IQ (ranging from 85 to 140) and age (college age, 60-74 years old, and 75-90 years old). IQ had large effects on drift rate in recognition and recall performance, except for the oldest participants with some measures near floor. Drift rates in the recognition tasks, accuracy in recall, and IQ all correlated strongly. However, there was a small decline in drift rates for item recognition and a large decline for associative recognition and cued recall accuracy (70%). In contrast, there were large effects of age on boundary separation and nondecision time (which correlated across tasks) but small effects of IQ. The implications of these results for single- and dual-process models of item recognition are discussed, and it is concluded that models that deal with both RTs and accuracy are subject to many more constraints than are models that deal with only one of these measures. Overall, the results of the study show a complicated but interpretable pattern of interactions that present important targets for modeling.     

Perceptual expertise improves category detection in natural scenes

There is much debate about how detection, categorization, and within-category identification relate to one another during object recognition. Whether these tasks rely on partially shared perceptual mechanisms may be determined by testing whether training on one of these tasks facilitates performance on another. In the present study we asked whether expertise in discriminating objects improves the detection of these objects in naturalistic scenes. Self-proclaimed car experts (N = 34) performed a car discrimination task to establish their level of expertise, followed by a visual search task where they were asked to detect cars and people in hundreds of photographs of natural scenes. Results revealed that expertise in discriminating cars was strongly correlated with car detection accuracy. This effect was specific to objects of expertise, as there was no influence of car expertise on person detection. These results indicate a close link between object discrimination and object detection performance, which we interpret as reflecting partially shared perceptual mechanisms and neural representations underlying these tasks: the increased sensitivity of the visual system for objects of expertise – as a result of extensive discrimination training – may benefit both the discrimination and the detection of these objects. Alternative interpretations are also discussed.     

Sex differences in memory for landmark arrays in C57BL/J6 mice

The most robust sex differences in cognition across polygynous mammalian species are the sex-specific patterns of the use of spatial cues during encoding and orientation. In laboratory rats, wild rodents, and humans, females orient preferentially to the features and arrangement of local landmarks, while males preferentially attend to distant landmarks. Yet this sex-specific pattern is often absent or reversed in the laboratory mouse, a species representing a major laboratory model of neural mechanisms. We explored sex differences in the C57BL/J6 strain of laboratory mouse by employing tasks that were motivated by the natural patterns of exploration. We predicted that such tasks would unmask the predicted default polygynous patterns of cue use by females and males. We used two standard tasks, a novel object recognition task and a five-stage serial object dishabituation task. On the first task, the results showed a female advantage in detecting the novel object, as predicted by prior results from other polygynous species. In the second task, we found, also as predicted, a male advantage in performance when the polarization of the array was distorted and a female advantage in performance when the local array was re-arranged. The pattern of sex-specific advantages in performance in C57BL/J6 mouse is thus concordant with that found in other polygynous mammals.     

What types of visual recognition tasks are mediated by the neural subsystem that subserves face recognition?

Three divided visual field experiments tested current hypotheses about the types of visual shape representation tasks that recruit the cognitive and neural mechanisms underlying face recognition. Experiment 1 found a right hemisphere advantage for subordinate but not basic-level face recognition. Experiment 2 found a right hemisphere advantage for basic but not superordinate-level animal recognition. Experiment 3 found that inverting animals eliminates the right hemisphere advantage for basic-level animal recognition. This pattern of results suggests that the cognitive and neural mechanisms underlying face recognition are recruited when computational demands of a shape representation task are best served through the use of coordinate (rather than categorical) spatial relations.     

Distributed representations of action sequences in anterior cingulate cortex: A recurrent neural network approach

Anterior cingulate cortex (ACC) has been the subject of intense debate over the past 2 decades, but its specific computational function remains controversial. Here we present a simple computational model of ACC that incorporates distributed representations across a network of interconnected processing units. Based on the proposal that ACC is concerned with the execution of extended, goal-directed action sequences, we trained a recurrent neural network to predict each successive step of several sequences associated with multiple tasks. In keeping with neurophysiological observations from nonhuman animals, the network yields distributed patterns of activity across ACC neurons that track the progression of each sequence, and in keeping with human neuroimaging data, the network produces discrepancy signals when any step of the sequence deviates from the predicted step. These simulations illustrate a novel approach for investigating ACC function.     

Hierarchical growing grid networks for skeleton based action recognition

In this paper, a novel cognitive architecture for action recognition is developed by applying layers of growing grid neural networks. Using these layers makes the system capable of automatically arranging its representational structure. In addition to the expansion of the neural map during the growth phase, the system is provided with a prior knowledge of the input space, which increases the processing speed of the learning phase. Apart from two layers of growing grid networks the architecture is composed of a preprocessing layer, an ordered vector representation layer and a one-layer supervised neural network. These layers are designed to solve the action recognition problem. The first-layer growing grid receives the input data of human actions and the neural map generates an action pattern vector representing each action sequence by connecting the elicited activation of the trained map. The pattern vectors are then sent to the ordered vector representation layer to build the time-invariant input vectors of key activations for the second-layer growing grid. The second-layer growing grid categorizes the input vectors to the corresponding action clusters/sub-clusters and finally the one-layer supervised neural network labels the shaped clusters with action labels. Three experiments using different datasets of actions show that the system is capable of learning to categorize the actions quickly and efficiently. The performance of the growing grid architecture is compared with the results from a system based on Self-Organizing Maps, showing that the growing grid architecture performs significantly superior on the action recognition tasks.     

An accurate and efficient multi-category edge detection method

Edge detection plays an important role in image processing. With the development of deep learning, the accuracy of edge detection has been greatly improved, and people have more requirements for edge detection tasks. Most edge detection algorithms are binary edge detection methods, but there are usually multiple categories of edges in an image. In this paper, we present an accurate multi-category edge detection network, the richer category-aware semantic edge detection network (R-CASENet). In order to make full use of convolutional neural network’s powerful feature expression capabilities, we attempt to use more information from feature maps for edge feature extraction and classification. Using the ResNet101 network as the backbone, firstly we merge the building blocks in different composite blocks and down-sample to obtain the feature maps. Then we fuse the feature maps in different composite blocks to obtain the final fused classifier. Experimental results show that R-CASENet can achieve state-of-the-art performance on the large SBD dataset. Furthermore, to get precise one-pixel width edges, we also propose an edge refinement network (ERN) structure. The proposed scheme is an end-to-end method and the proposed ERN can reduce redundant points and improve computational efficiency, especially for further image processing.