Phonological i-Vectors to Detect Parkinson’s Disease

A central issue in speech recognition is which basic units of speech are extracted by the auditory system and used for lexical access. One suggestion is that complex acoustic-phonetic information is mapped onto abstract phonological representations of speech and that a finite set of phonological features is used to guide speech perception. Previous studies analyzing the N1m component of the auditory evoked field have shown that this holds for the acoustically simple feature place of articulation. Brain magnetic correlates indexing the extraction of acoustically more complex features, such as lip rounding (ROUND) in vowels, have not been unraveled yet. The present study uses magnetoencephalography (MEG) to describe the spatial-temporal neural dynamics underlying the extraction of phonological features. We examined the induced electromagnetic brain response to German vowels and found the event-related desynchronization in the upper beta-band to be prolonged for those vowels that exhibit the lip rounding feature (ROUND). It was the presence of that feature rather than circumscribed single acoustic parameters, such as their formant frequencies, which explained the differences between the experimental conditions. We conclude that the prolonged event-related desynchronization in the upper beta-band correlates with the computational effort for the extraction of acoustically complex phonological features from the speech signal. The results provide an additional biomagnetic parameter to study mechanisms of speech perception.

The work is devoted to resolving the contradiction between the accuracy of modelling nonlinear dynamics and the speed of model construction under conditions of limited computing resources. The purposeof the work is to reduce the time for building time delay neural networks while ensuring a given accuracy in the tasks of identifying nonlinear dynamic objects with continuous characteristics. This goal is achieved by developing a method for pre-training neural networks that reflect the basic characteristics of the subject area. The scientific novelty of the work is the development of a method for identifying nonlinear dynamic objectsin the form of time delay neural networks based on the use of a set of basic pre-trained neural networks that reflect the typical properties of the subject area. In contrast to the traditional approach to pre-training, the developed method allows building models of lower complexity.A formal criterion is proposed for determining the moment of termination of the neural network pre-training, the use of which allows avoiding retraining of the base model and ensuring a significant reduction in the model training time on the target data set. The practical utility of the work lies in the development of an algorithm for the method of pre-training time delay neural networks in the tasks of identifying nonlinear dynamic objects with continuous characteristics, which allows to significantlyreduce the training time of neural networks without losing model accuracy. The value of this study is todetermine the area of effective use of the proposed method, namely, when the general and target datasets do not have significant differences and the target dataset is of sufficient size to reflect the properties of the research object.

Neural style transfer has recently received significant attention and demonstrated amazing results. An efficient solution proposed by Johnson et al. trains feed-forward convolutional neural networks by defining and optimizing perceptual loss functions. Such methods are typically based on high-level features extracted from pre-trained neural networks, where the loss functions contain two components: style loss and content loss. However, such pre-trained networks are originally designed for object recognition, and hence the high-level features often focus on the primary target and neglect other details. As a result, when input images contain multiple objects potentially at different depths, the resulting images are often unsatisfactory because image layout is destroyed and the boundary between the foreground and background as well as different objects becomes obscured. We observe that the depth map effectively reflects the spatial distribution in an image and preserving the depth map of the content image after stylization helps produce an image that preserves its semantic content. In this paper, we introduce a novel approach for neural style transfer that integrates depth preservation as additional loss, preserving overall image layout while performing style transfer.

BackgroundDetection of ossification areas of hand bones in X-ray images is an important task, e.g. as a preprocessing step in automated bone age estimation. Deep neural networks have emerged recently as de facto standard detection methods, but their drawback is the need of large annotated datasets. Finetuning pre-trained networks is a viable alternative, but it is not clear a priori if training with small annotated datasets will be successful, as it depends on the problem at hand. In this paper, we show that pre-trained networks can be utilized to produce an effective detector of ossification areas in pediatric X-ray images of hands.Methods and findingsA publicly available Faster R-CNN network, pre-trained on the COCO dataset, was utilized and finetuned with 240 manually annotated radiographs from the RSNA Pediatric Bone Age Challenge, which comprises over 14.000 pediatric radiographs. The validation is done on another 89 radiographs from the dataset and the performance is measured by Intersection-over-Union (IoU). To understand the effect of the data size on the pre-trained network, subsampling was applied to the training data and the training was repeated. Additionally, the network was trained from scratch without any pre-trained weights. Finally, to understand whether the trained model could be useful, we compared the inference of the network to an annotation of an expert radiologist. The finetuned network was able to achieve an average precision (mAP@0.5IoU) of 92.92 ± 1.93. Apart from the wrist region, all ossification areas were able to benefit from more data. In contrast, the network trained from scratch was not able to produce any correct results. When compared to the annotations of the expert radiologist, the network was able to localize the regions quite well, as the F1-Score was on average 91.85 ± 1.06.ConclusionsBy finetuning a pre-trained deep neural network, with 240 annotated radiographs, we were able to successfully detect ossification areas in prediatric hand radiographs.

Advances in satellite imagery presents unprecedented opportunities for understanding natural and social phenomena at global and regional scales. Although the field of satellite remote sensing has evaluated imperative questions to human and environmental sustainability, scaling those techniques to very high spatial resolutions at regional scales remains a challenge. Satellite imagery is now more accessible with greater spatial, spectral and temporal resolution creating a data bottleneck in identifying the content of images. Because satellite images are unlabeled, unsupervised methods allow us to organize images into coherent groups or clusters. However, the performance of unsupervised methods, like all other machine learning methods, depends on features. Recent studies using features from pre-trained networks have shown promise for learning in new datasets. This suggests that features from pre-trained networks can be used for learning in temporally and spatially dynamic data sources such as satellite imagery. It is not clear, however, which features from which layer and network architecture should be used for learning new tasks. In this paper, we present an approach to evaluate the transferability of features from pre-trained Deep Convolutional Neural Networks for satellite imagery. We explore and evaluate different features and feature combinations extracted from various deep network architectures, and systematically evaluate over 2,000 network-layer combinations. In addition, we test the transferability of our engineered features and learned features from an unlabeled dataset to a different labeled dataset. Our feature engineering and learning are done on the unlabeled Draper Satellite Chronology dataset, and we test on the labeled UC Merced Land dataset to achieve near state-of-the-art classification results. These results suggest that even without any or minimal training, these networks can generalize well to other datasets. This method could be useful in the task of clustering unlabeled images and other unsupervised machine learning tasks.

enhancement is a task to improve the intelligibility and perceptual quality of degraded speech signal. Recently, neural networks based methods have been applied to speech enhancement. However, many neural network based methods require noisy and clean speech pairs for training. We propose a speech enhancement framework that can be trained with large-scale weakly labelled AudioSet dataset. Weakly labelled data only contain audio tags of audio clips, but not the onset or offset times of speech. We first apply pretrained audio neural networks (PANNs) to detect anchor segments that contain speech or sound events in audio clips. Then, we randomly mix two detected anchor segments containing speech and sound events as a mixture, and build a conditional source separation network using PANNs predictions as soft conditions for speech enhancement. In inference, we input a noisy speech signal with the one-hot encoding of Speech as a condition to the trained system to predict enhanced speech. Our system achieves a PESQ of 2.28 and an SSNR of 8.75 dB on the VoiceBank-DEMAND dataset, outperforming the previous SEGAN system of 2.16 and 7.73 dB respectively.

Measurements were made of distal radius, mid-radius, tibia and metatarsal bone-density of 80 patients with myelomeningocele (17 thoracic, six L1/L2, 13 L3, 30 L4, 14 L5/sacral). For the upper extremity the bone density primarily was low in the thoracic patients, but in the tibia and metatarsal it showed a more linear correlation with neurological levels. The effect of age was highly significant at all sites; after controlling for this, the neurological level was a significant determinant of bone density at all sites, and this effect was greater in older children. Patients with impaired ambulation had decreased bone-density in the distal radius, tibia and metatarsal, but not in the mid-radius. Race had no significant effect on density after accounting for differences in neurological level. Weight for height and multiple fractures did not correlate with bone density. Although ambulatory status (weight-bearing stresses) and neurological status (muscle stresses) are both important factors in bone density, this study suggests that the latter is a more important determinant.

A computer-based intelligent visual inspection system plays a major role in evaluating the quality of textile fabrics and its demand is continuously increasing in the textile industry, especially when the quality of textile is to be considered. In this paper, we propose an AI-based automated fabric defect detection algorithm which utilizes pre-trained deep neural network models for classifying possible fabric defects. The fabric images are enhanced by pre-processing at various levels using conventional image processing techniques and they are used to train the networks. The Deep Convolutional Neural Network (DCNN) and a pre-trained network, AlexNet, are used to train and classify various fabric defects. With the exiting textile dataset, a maximum classification accuracy of 92.60% is achieved in the conducted simulations. With this accuracy, the detection and classification system based on this classifier model can aid the human to find faults in the fabric manufacturing unit.

We present a simple technique that can convert a pretrained segmentation neural network to a salient object detector. We show that the pretrained network can be agnostic to the semantic class of the object of interest, and no further training is required. Experiments were run on UAV-captured aerial imagery of the “smart home” structure located in the premises of the CERTH research center. Further experiments were also run on natural scenes. Our tests validate the usefulness of the proposed technique.

Automated detection of leukemia by pretrained deep neural networks and transfer learning: A comparison

Neurological level effect on the discharge functional status of spinal cord injured persons after rehabilitation

The process of speech production is very complex because conceptual content must be transformed in a language form and articulated in a physical, primarily acoustic signal which can activate sensory abilities of a listener, who will use the physical speech signal to discover the meaning initiated by the entire communi cative process. Since the productive and perceptual abilities of people are coordi nated and enable an efficient communication and exchange of information, the speaker tries to produce a fluent speech. However, because of the complexity of speech communication and biological imperfection of humans, it is only natural for speech to sometimes contain certain disfluencies. The non-phonemic sound /ǝ/ is one such example. This research encompassed 20 subjects who were inter viewed on the Zagreb Radio’s Prvi program [First Programme] and Omladinski radio [Youth Radio]. The analysis contained 200 seconds of speech per partici pant, deemed as formal in Prvi program and informal in Omladinski radio, and was conducted through the Praat computer programme for speech analysis. The research indicated a significant difference in the number of occurrences of the non-phonemic sound, appearing 147 times in formal and 330 times in informal speech. The average duration of the sound production shows no significant dif ference: 0.354 s in formal and 0.316 s in informal speech. The difference in the number of occurrences was statistically significant and the difference in the aver age duration was not.

Background:Pain is a common symptom in Parkinson's disease (PD) patients. Scales to rate pain in PD are marred by several flaws, either not being available in other languages or not specific for PD.Objectives:To assess the frequency of pain among bilingual Indian PD patients using “King's Parkinson's disease pain scale” (KPPS) and to validate it.Methods:We randomly administered KPPS in Hindi/English to all consecutive bilingual persons with PD. The results were appropriately analyzed.Results:A total of 119 PD patients were enrolled with a mean age of 64.34 (± 9.57) years. Median Hoehn and Yahr stage was 2 (42.85%). Pain was present in 62 (52.1%) PD patients. The most common type was musculoskeletal (74.19%). The mean total KPPS score was 16.02 ± 10.57. KPPS score was significantly higher in women and correlated positively with unified Parkinson's disease rating scale (UPDRS) part 2 and 4 scores (r = 0.27 and r = 0.25). Risk factors for pain were female gender, higher H and Y stage, total UPDRS score, and individual UPDRS part 3 and 4 scores. Difficulty falling asleep (P = 0.01), frequent awakenings (P = 0.01), diminished smell sensation (P = 0.003), diminished speech volume (P = 0.02), gait freezing (P = 0.03), and falls (P = 0.001) correlated with the presence of pain. The interclass correlation coefficient between the Hindi and English versions of KPPS was 0.835, while Bland–Altman analysis showed 96.7% agreement suggesting excellent correlation and validation.Conclusions:KPPS is an easy tool for characterization, scoring, and follow-up of pain in PD patients. The Hindi version has good agreement with the original English version.

Routine inspection of a power line insulator system for early problem detection and maintenance is required for the effective transmission of electrical power to customers. Unlike the traditional manual inspection methods such as foot patrol, helicopter assisted methods, flying and climbing robots are time consuming, dangerous, labor intensive, and expensive. This work primarily focused on the insulator image classification of the Chinese Power Line Insulator Dataset (CPLID). Convolutional Neural Network (CNN) and pre-trained neural networks with different optimizers like Adam, Adamax, Adagrad, Adadelta, Nadam, Ftrl, and RMSprop were used to find out the best combination of CNN and pre-trained neural network with the optimizer to provide robust performance and higher accuracy. Finally, CNN with Adamax optimizer provided overall accuracy of 95.27% and the pre-trained CNN Densenet 201 with Adam optimizer gave overall accuracy of 100%. In the automated inspection of transmission line insulators, the deployment of UAVs and deep learning algorithms can be a useful starting point for many researchers throughout the world.

Pre-trained single-channel neural networks have become more prevalent for noise reduction in recent years. However, unlike their multichannel counterparts, these monoaural approaches do not exploit spatial information during the optimization process. Furthermore, while multichannel neural networks exploit spatial information, they are optimized for a specific microphone array configuration; extensive data collection and training are required if a new array configuration is deployed. We propose a transfer learning approach that leverages existing pre-trained single-channel neural networks for the optimization of multichannel neural networks. Simulation results on the CHiME-3 dataset show that the proposed method outperforms the state-of-the-art multichannel neural network and neural beamformer.