Biometric Recognition Performing in a Bioinspired System

In this study, we propose a set of biometric recognition experiments in similar conditions to real operating systems. This implies a jump from the usual laboratory conditions to a more real situation where the amount of variability between training and testing samples is large. We present experiments with face, hand-geometry, and signature recognition training a “universal classifier” able to decide if two input samples belong to the same person or not. During test, we recognize samples of a different database not used during classifier training. Training with the ORL face database and testing with the AR database provides a 5.1% error rate in verification operation, while training and testing with the same database yields 2.5%. For hand-geometry databases, we obtain 4.33 and 0.16% for different and same testing and training databases, respectively. For signature recognition, we obtain 1.36 and 4.14% for different and same testing and training databases, respectively. Our proposed system implies a very low computational cost to introduce/remove a user in the database, which is a crucial point for a real operation biometric system.     

Electrophysiological evidence for prelinguistic infants' word recognition in continuous speech

Children begin to talk at about age one. The vocabulary they need to do so must be built on perceptual evidence and, indeed, infants begin to recognize spoken words long before they talk. Most of the utterances infants hear, however, are continuous, without pauses between words, so constructing a vocabulary requires them to decompose continuous speech in order to extract the individual words. Here, we present electrophysiological evidence that 10-month-old infants recognize two-syllable words they have previously heard only in isolation when these words are presented anew in continuous speech. Moreover, they only need roughly the first syllable of the word to begin doing this. Thus, prelinguistic infants command a highly efficient procedure for segmentation and recognition of spoken words in the absence of an existing vocabulary, allowing them to tackle effectively the problem of bootstrapping a lexicon out of the highly variable, continuous speech signals in their environment.     

Self partitioning neural networks for target recognition

Automatic target recognition (ATR) is a domain in which the neural network technology has been applied with limited success. The domain is characterized by large training sets with dissimilar target images carrying conflicting information. This paper presents a novel method for quantifying the degree of non-cooperation that exists among the target members of the training set. Both the network architecture and the training algorithm are considered in the computation of the non-cooperation measures. Based on these measures, the self partitioning neural network (SPNN) approach partitions the target vectors into an appropriate number of groups and trains one subnetwork to recognize the targets in each group. A fusion network combines the outputs of the subnetworks to produce the final response. This method automatically determines the number of subnetworks needed without excessive computation. The subnetworks are simple with only one hidden layer and one unit in the output layer. They are topologically identical to one another. The simulation results indicate that the method is robust and capable of self organization to overcome the ill effects of the non-cooperating targets in the training set. The self partitioning approach improves the classification accuracy and reduces the training time of neural networks significantly. It is also shown that a trained self partitioning neural network is capable of learning new training vectors without retraining on the combined training set (i.e., the training set consisting of the previous and newly acquired training vectors).     

Experiments for isolated-word recognition with single- and two-layer perceptrons

Several design strategies for feed-forward networks are examined within the scope of pattern classification. Single- and two-layer perceptron models are adapted for experiments in isolated-word recognition. Direct (one-step) classification as well as several hierarchical (two-step) schemes have been considered. For a vocabulary of 20 English words spoken repeatedly by 11 speakers, the word classes are found to be separable by hyperplanes in the chosen feature space. Since for speaker-dependent word recognition the underlying data base contains only a small training set, an automatic expansion of the training material improves the generalization properties of the networks. This method accounts for a wide variety of observable temporal structures for each word and gives a better overall estimate of the network parameters which leads to a recognition rate of 99.5%. For speaker-independent word recognition, a hierarchical structure with pairwise training of two-class models is superior to a single uniform network (98% average recognition rate).     

Enhancing the ability of adults with mental retardation to recognize facial expressions of emotion.

The ability to recognize accurately and respond appropriately to facial expressions of emotion is essential for interpersonal interaction. Individuals with mental retardation typically are deficient in these skills. The ability of 7 adults, 1 with severe and 6 with moderate mental retardation, to recognize facial expressions of emotion correctly was assessed. Then, they were taught this skill using a combination of a discrimination training procedure for differentiating facial movements, directed rehearsal, and Ekman and Friesen's "flashing photograph" technique. Their average increase in accuracy over baseline was at least 30% during the course of the training and over 50% during the last 5 days of the training phase. Further, these individuals were able to generalize their skills from posed photographs to videotaped role plays and were able to maintain their enhanced skills during the 8 to 9 months following the termination of training. This is the first study to show that individuals with mental retardation can be taught skills that enhance their ability to recognize facial expressions of emotion.     

Training multi-layered neural network neocognitron

This paper proposes new learning rules suited for training multi-layered neural networks and applies them to the neocognitron. The neocognitron is a hierarchical multi-layered neural network capable of robust visual pattern recognition. It acquires the ability to recognize visual patterns through learning. For training intermediate layers of the hierarchical network of the neocognitron, we use a new learning rule named add-if-silent. By the use of the add-if-silent rule, the learning process becomes much simpler and more stable, and the computational cost for learning is largely reduced. Nevertheless, a high recognition rate can be kept without increasing the scale of the network. For the highest stage of the network, we use the method of interpolating-vector. We have previously reported that the recognition rate is greatly increased if this method is used during recognition. This paper proposes a new method of using it for both learning and recognition. Computer simulation demonstrates that the new neocognitron, which uses the add-if-silent and the interpolating-vector, produces a higher recognition rate for handwritten digits recognition with a smaller scale of the network than the neocognitron of previous versions.     

Evolving spiking neural networks for audiovisual information processing

This paper presents a new modular and integrative sensory information system inspired by the way the brain performs information processing, in particular, pattern recognition. Spiking neural networks are used to model human-like visual and auditory pathways. This bimodal system is trained to perform the specific task of person authentication. The two unimodal systems are individually tuned and trained to recognize faces and speech signals from spoken utterances, respectively. New learning procedures are designed to operate in an online evolvable and adaptive way. Several ways of modelling sensory integration using spiking neural network architectures are suggested and evaluated in computer experiments.     

Dissociation between face perception and face memory in adults,but not children,with developmental prosopagnosia

Cognitive models propose that face recognition is accomplished through a series of discrete stages, including perceptual representation of facial structure, and encoding and retrieval of facial information. This implies that impaired face recognition can result from failures of face perception, face memory, or both. Studies of acquired prosopagnosia, autism spectrum disorders, and the development of normal face recognition support the idea that face perception and face memory are distinct processes, yet this distinction has received little attention in developmental prosopagnosia (DP). To address this issue, we tested the face perception and face memory of children and adults with DP. By definition, face memory is impaired in DP, so memory deficits were present in all participants. However, we found that all children, but only half of the adults had impaired face perception. Thus, results from adults indicate that face perception and face memory are dissociable, while the results from children provide no evidence for this division. Importantly, our findings raise the possibility that DP is qualitatively different in childhood versus adulthood. We discuss theoretical explanations for this developmental pattern and conclude that longitudinal studies are necessary to better understand the developmental trajectory of face perception and face memory deficits in DP.     

Pattern recognition of the electroencephalogram by artificial neural networks

A back-propagation network was trained to recognize high voltage spike-wave spindle (HVS) patterns in the rat, a rodent model of human petit mal epilepsy. The spontaneously occurring HVSs were examined in 137 rats of the Fisher 344 and Brown Norway strains and their F1, F2 and backcross hybrids. Neocortical EEG and movement of the rat were recorded for 12 night hours in each animal and analog data were filtered (low cut: 1 Hz; high cut: 50 Hz) and sampled at 100 Hz with 12 bit precision. A training data set was generated by manually marking durations of HVS epochs in 16 representative animals selected from each group. Training data were presented to back-propagation networks with variable numbers of input, hidden and output cells. The performance of different types of networks was first examined with the training samples and then the best configuration was tested on novel sets of the EEG data. FFT transformation of EEG significantly improved the pattern recognition ability of the network. With the most effective configuration (16 input; 19 hidden; 1 output cells) the summed squared error dropped by 80% as compared with that of the initial random weights. When testing the network with new patterns the manual and automatic evaluations were compared quantitatively. HVSs which were detected properly by the network reached 93–99% of the manually marked HVS patterns, while falsely detected events (non-HVS, artifacts) varied between 18% and 40%. These findings demonstrate the utility of back-propagation networks in automatic recognition of EEG patterns.     

A Class Incremental Extreme Learning Machine for Activity Recognition

Automatic activity recognition is an important problem in cognitive systems. Mobile phone-based activity recognition is an attractive research topic because it is unobtrusive. There are many activity recognition models that can infer a user’s activity from sensor data. However, most of them lack class incremental learning abilities. That is, the trained models can only recognize activities that were included in the training phase, and new activities cannot be added in a follow-up phase. We propose a class incremental extreme learning machine (CIELM). It (1) builds an activity recognition model from labeled samples using an extreme learning machine algorithm without iterations; (2) adds new output nodes that correspond to new activities; and (3) only requires labeled samples of new activities and not previously used training data. We have tested the method using activity data. Our results demonstrated that the CIELM algorithm is stable and can achieve a similar recognition accuracy to the batch learning method.     

Extending the evaluation of a computer system used as a microswitch for word utterances of persons with multiple disabilities

Background Microswitches can be vital tools to help individuals with extensive multiple disabilities acquire control of environmental stimulation. This study was aimed at extending the evaluation of a computer system used as a microswitch for word utterances with three participants with multiple disabilities. Method Sets of 7 or 12 word utterances were used for the participants. The utterances were divided into three groups, which were exposed to intervention successively. During the intervention and a 2-month post-intervention check, the participants’ emission of the target utterances led the system to present favourite, matching stimuli (i.e. provided that it recognized the utterances). Results Intervention data showed that (1) the participants increased the frequencies of the target utterances and (2) the computer system recognized approximately 80% of those utterances. These findings were maintained at the post-intervention check. An analysis of the levels of occurrence of individual utterances showed statistically significant differences among them, in line with the notions of preference and choice. Conclusions The computer system was useful as a microswitch to enable access to favourite stimuli. There is a need to improve the accuracy of the system with respect to its recognition of the participants’ utterances.     

Morphometric analysis of peripheral myelinated nerve fibers through deep learning

Irrespective of initial causes of neurological diseases, these disorders usually exhibit two key pathological changes—axonal loss or demyelination or a mixture of the two. Therefore, vigorous quantification of myelin and axons is essential in studying these diseases. However, the process of quantification has been labor intensive and time‐consuming because of the requisite manual segmentation of myelin and axons from microscopic nerve images. As a part of AI development, deep learning has been utilized to automate certain tasks, such as image analysis. This study describes the development of a convolutional neural network (CNN)—based approach to segment images of mouse nerve cross sections. We adapted the U‐Net architecture and used manually‐produced segmentation data accumulated over many years in our lab for training. These images ranged from normal nerves to those afflicted by severe myelin and axon pathologies; thus, maximizing the trained model's ability to recognize atypical myelin structures. Morphometric data produced by applying the trained model to additional images were then compared to manually obtained morphometrics. The former effectively shortened the time consumption in the morphometric analysis with excellent accuracy in axonal density and g‐ratio. However, we were not able to completely eliminate manual refinement of the segmentation product. We also observed small variations in axon diameter and myelin thickness within 9.5%. Nevertheless, we learned alternative ways to improve accuracy through the study. Overall, greatly increased efficiency in the CNN‐based approach out‐weighs minor limitations that will be addressed in future studies, thus justifying our confidence in its prospects. Note: All the relevant code is freely available at https://neurology.med.wayne.edu/drli‐datashairing     

ZFNet模型在胶质瘤MRI诊断中的应用

研究背景胶质瘤疾病负担较重,尽早确诊和及时治疗可有效延长无进展生存期,临床实践中初诊疑似胶质瘤时首选头部MRI检查,人工阅片存在诊断结果不一致和阅片效率下降的缺陷,而通过深度学习算法进行医学影像识别与诊断成为可能。本研究采用人工神经网络相关机器学习算法,辅助影像科医师对胶质瘤患者头部MRI图像的人工阅片,以期改善人工阅片耗时、费力以及因主观判断导致阅片结果不同的缺陷。方法纳入TCIA数据库中130例成年胶质瘤患者计40036张头部MRI图像,随机分为训练集(28025张)和测试集(12011张),再根据医学专家的标注定义为“肿瘤影像”和“正常影像”,采用ZFNet模型进行图像识别与分类模型的建立,绘制训练集的强化学习曲线,观察训练准确度随训练步数变化的趋势。将测试集导入模型,计算ZFNet模型预测“肿瘤影像”的分类准确率、阳性预测值、灵敏度、特异度和F1值。同时进行AlexNet模型对比建模,与ZFNet模型结果进行比较。结果ZFNet模型在训练38757步后训练准确度稳定为99.7%,AlexNet模型则在训练37984步后稳定为98.23%;将测试集导入ZFNet模型,ZFNet模型预测“肿瘤影像”的准确度为84.42%(10140/12011)、阳性预测值为80.77%(4817/5964)、灵敏度为86.93%(4817/5541)、特异度为82.27%(5323/6470)、F1值为83.74%,AlexNet模型为80.74%(9698/12011)、77.68%(4529/5830)、81.74%(4529/5541)、79.89%(5169/6470)和79.66%,ZFNet模型在各个维度的分类性能均优于AlexNet模型,效果满意。结论ZFNet模型在胶质瘤患者头部MRI图像分类预测方面的效果尚佳,可为建立胶质瘤影像学辅助诊断模型提供良好的技术支持。     

Clinical recognition of dementia and cognitive impairment in primary care: a meta-analysis of physician accuracy

Mitchell AJ, Meader N, Pentzek M. Clinical recognition of dementia and cognitive impairment in primary care: a meta‐analysis of physician accuracy. Objective: We aimed to examine the ability of the general practitioners (GPs) to recognize a spectrum of cognitive impairment from mild cognitive impairment (MCI) to severe dementia in routine practice using their own clinical judgment. Method: Using PRISMA criteria, a meta‐analysis of studies testing clinical judgment and clinical documentation was conducted against semi‐structured interviews (for dementia) and cognitive tests (for cognitive impairment). We located 15 studies reporting on dementia, seven studies that examined recognition of broadly defined cognitive impairment, and eight regarding MCI. Results: By clinical judgment, clinicians were able to identify 73.4% of people with dementia and 75.5% of those without dementia but they made correct annotations in medical records in only 37.9% of cases (and 90.5% of non‐cases). For cognitive impairment, detection sensitivity was 62.8% by clinician judgment but 33.1% according to medical records. Specificity was 92.6% for those without cognitive impairment by clinical judgment. Regarding MCI, GPs recognized 44.7% of people with MCI, although this was recorded in medical notes only 10.9% of the time. Their ability to identify healthy individuals without MCI was between 87.3% and 95.5% (detection specificity). Conclusion: GPs have considerable difficulty identifying those with MCI and those with mild dementia and are generally poor at recording such diagnoses in medical records.     

Feature Component-Based Extreme Learning Machines for Finger Vein Recognition

This paper proposes an efficient finger vein recognition system, in which a variant of the original ensemble extreme learning machine (ELM) called the feature component-based ELMs (FC-ELMs) designed to utilize the characteristics of the features, is introduced to improve the recognition accuracy and stability and to substantially reduce the number of hidden nodes. For feature extraction, an explicit guided filter is proposed to extract the eight block-based directional features from the high-quality finger vein contours obtained from noisy, non-uniform, low-contrast finger vein images without introducing any segmentation process. An FC-ELMs consist of eight single ELMs, each trained with a block feature with a pre-defined direction to enhance the robustness against variation of the finger vein images, and an output layer to combine the outputs of the eight ELMs. For the structured training of the vein patterns, the FC-ELMs are designed to first train small differences between patterns with the same angle and then to aggregate the differences at the output layer. Each ELM can easily learn lower-complexity patterns with a smaller network and the matching accuracy can also be improved, due to the less complex boundaries required for each ELM. We also designed the ensemble FC-ELMs to provide the matching system with stability. For the dataset considered, the experimental results show that the proposed system is able to generate clearer vein contours and has good matching performance with an accuracy of 99.53 % and speed of 0.87 ms per image.     

Epileptic EEG classification based on extreme learning machine and nonlinear features

The automatic detection and classification of epileptic EEG are significant in the evaluation of patients with epilepsy. This paper presents a new EEG classification approach based on the extreme learning machine (ELM) and nonlinear dynamical features. The theory of nonlinear dynamics has been a powerful tool for understanding brain electrical activities. Nonlinear features extracted from EEG signals such as approximate entropy (ApEn), Hurst exponent and scaling exponent obtained with detrended fluctuation analysis (DFA) are employed to characterize interictal and ictal EEGs. The statistics indicate that the differences of those nonlinear features between interictal and ictal EEGs are statistically significant. The ELM algorithm is employed to train a single hidden layer feedforward neural network (SLFN) with EEG nonlinear features. The experiments demonstrate that compared with the backpropagation (BP) algorithm and support vector machine (SVM), the performance of the ELM is better in terms of training time and classification accuracy which achieves a satisfying recognition accuracy of 96.5% for interictal and ictal EEG signals.     

Reading aloud as rehabilitation method for children with dyslexia detected at the first grade in their primary school

We provided reading aloud instructions to a child who was diagnosed with dyslexia in a regular class of 69 first graders, comprising 33 boys and 36 girls, during a test of reading sentences aloud. The instructions consisted of a 2-step approach, i.e., decoding instruction and vocabulary instruction. First, a decoding instruction, which emphasized an important point in effortless decoding, was presented to the child. Next, a vocabulary instruction, which aimed to facilitate word-form recognition, was provided. We found that, the decoding instruction was effective in decreasing the number of reading errors, and that the vocabulary instruction was effective against reducing the time taken to read aloud.     

Fat neural network for recognition of position-normalised objects

The design of a recognition system for natural objects is difficult, mainly because such objects are subject to a strong variability that cannot be easily modelled: planktonic species possess such highly variable forms. Existing plankton recognition systems usually comprise feature extraction processing upstream of a classifier. Drawbacks of such an approach are that the design of relevant feature extraction processes may be very difficult, especially if classes are numerous and if intra-class variability is high, so that the system becomes specific to the problem for which features have been tuned. The opposite course that we take is based on a structured multi-layer neural network with no shared weights, which generates its own features during training. Such a large parameterised—fat—network exhibits good generalisation capabilities for pattern recognition problems dealing with position-normalised objects, even with as many as one thousand weights as training examples. The advantage of such large networks, in terms of generalisation efficiency, adaptability and classification time, is demonstrated by applying the network to three plankton recognition and face recognition problems. Its ability to perform good generalisation with few training examples, but many weights, is an open theoretical problem.