Dataset of dust mass accumulation rates for the loess-palaeosol sequences from the Carpathian Basin.

In this article, a dataset of age-depth modelling data, sedimentation rates and dust mass accumulation rates (MAR) from four loess-palaeosol sequences from the Carpathian Basin is presented. The dataset is related to the article "Detailed luminescence dating of dust mass accumulation rates over the last two glacial-interglacial cycles from the Irig loess-palaeosol sequence, Carpathian Basin", published in the journal Global and Planetary Change by Peric et al. [1]. In the dataset, luminescence ages from the loess sites Irig, Nosak, Stari Slankamen and Crvenka were modeled using the r.bacon software after which the dust mass accumulation rates were calculated. For a more realistic representation the MARs were subsequently smoothed using the SigmaPlot software. For all sites, minimum, maximum, median and mean values for the modelled ages and accumulation rates are calculated and presented.

Speaker Recognition Using Constrained Convolutional Neural Networks in Emotional Speech.

Speaker recognition is an important classification task, which can be solved using several approaches. Although building a speaker recognition model on a closed set of speakers under neutral speaking conditions is a well-researched task and there are solutions that provide excellent performance, the classification accuracy of developed models significantly decreases when applying them to emotional speech or in the presence of interference. Furthermore, deep models may require a large number of parameters, so constrained solutions are desirable in order to implement them on edge devices in the Internet of Things systems for real-time detection. The aim of this paper is to propose a simple and constrained convolutional neural network for speaker recognition tasks and to examine its robustness for recognition in emotional speech conditions. We examine three quantization methods for developing a constrained network: floating-point eight format, ternary scalar quantization, and binary scalar quantization. The results are demonstrated on the recently recorded SEAC dataset.

Whether the Support Region of Three-Bit Uniform Quantizer Has a Strong Impact on Post-Training Quantization for MNIST Dataset?

Driven by the need for the compression of weights in neural networks (NNs), which is especially beneficial for edge devices with a constrained resource, and by the need to utilize the simplest possible quantization model, in this paper, we study the performance of three-bit post-training uniform quantization. The goal is to put various choices of the key parameter of the quantizer in question (support region threshold) in one place and provide a detailed overview of this choice's impact on the performance of post-training quantization for the MNIST dataset. Specifically, we analyze whether it is possible to preserve the accuracy of the two NN models (MLP and CNN) to a great extent with the very simple three-bit uniform quantizer, regardless of the choice of the key parameter. Moreover, our goal is to answer the question of whether it is of the utmost importance in post-training three-bit uniform quantization, as it is in quantization, to determine the optimal support region threshold value of the quantizer to achieve some predefined accuracy of the quantized neural network (QNN). The results show that the choice of the support region threshold value of the three-bit uniform quantizer does not have such a strong impact on the accuracy of the QNNs, which is not the case with two-bit uniform post-training quantization, when applied in MLP for the same classification task. Accordingly, one can anticipate that due to this special property, the post-training quantization model in question can be greatly exploited.

Design of a 2-Bit Neural Network Quantizer for Laplacian Source.

Achieving real-time inference is one of the major issues in contemporary neural network applications, as complex algorithms are frequently being deployed to mobile devices that have constrained storage and computing power. Moving from a full-precision neural network model to a lower representation by applying quantization techniques is a popular approach to facilitate this issue. Here, we analyze in detail and design a 2-bit uniform quantization model for Laplacian source due to its significance in terms of implementation simplicity, which further leads to a shorter processing time and faster inference. The results show that it is possible to achieve high classification accuracy (more than 96% in the case of MLP and more than 98% in the case of CNN) by implementing the proposed model, which is competitive to the performance of the other quantization solutions with almost optimal precision.

Speech Technology Progress Based on New Machine Learning Paradigm.

Speech technologies have been developed for decades as a typical signal processing area, while the last decade has brought a huge progress based on new machine learning paradigms. Owing not only to their intrinsic complexity but also to their relation with cognitive sciences, speech technologies are now viewed as a prime example of interdisciplinary knowledge area. This review article on speech signal analysis and processing, corresponding machine learning algorithms, and applied computational intelligence aims to give an insight into several fields, covering speech production and auditory perception, cognitive aspects of speech communication and language understanding, both speech recognition and text-to-speech synthesis in more details, and consequently the main directions in development of spoken dialogue systems. Additionally, the article discusses the concepts and recent advances in speech signal compression, coding, and transmission, including cognitive speech coding. To conclude, the main intention of this article is to highlight recent achievements and challenges based on new machine learning paradigms that, over the last decade, had an immense impact in the field of speech signal processing.

Outcome prediction of initial lamotrigine monotherapy in adult patients with newly diagnosed localization related epilepsies.

OBJECTIVE: To develop and test model to predict outcome of treatment with initial lamotrigine monotherapy in adult patients with newly diagnosed localization - related epilepsy, using data available at the time of diagnosis. METHODS: Prospective longitudinal study included consecutive series of adult patients with newly diagnosed localization - related epilepsy started of lamotrigine monotherapy. Logistic regression analysis using backward procedure was performed with treatment failure as the outcome variable. We evaluated both calibration and discrimination of the model. Internal validation of the model was performed with bootstrapping techniques. RESULTS: A total of 159 patients on lamotrigine monotherapy have been included in final analysis. Among them 78 (49.06%) patients had persistent seizures. Finally fitted multivariate model included: 1) age at therapy start, 2) presence of complex partial seizures, 3) aetiology of epilepsy and 4) interaction of age and epilepsy aetiology. Estimated odds ratio for seizure relapse in old patients with symptomatic epilepsy is lower than for the old patients with cryptogenic epilepsy, despite strong positive covariate effect of epilepsy aetiology. The model correctly classified 69.23% patients with seizure relapses and 81.48% of patients with seizure freedom, with estimated c - statistic of 0.80. Testing practical application we observed threefold increase or reduction of odds for the seizure relapse after model's positive or negative prediction respectively. CONCLUSION: Standard clinical data were modesty adequate to predict response to the initial trial of lamotrigine in adult patients with localization related epilepsy. Better markers of antiepileptic failure are required to guide optimal patient counselling and clinical decisions. Formal interaction analysis of variables improves outcome prediction and may be a key to correct interpretation of data.

Artificial neural networks based early clinical prediction of mortality after spontaneous intracerebral hemorrhage.

Numerous outcome prediction models have been developed for mortality and functional outcome after spontaneous intracerebral haemorrhage (ICH). However, no outcome prediction model for ICH has considered the impact of care restriction. To develop and compare results of the artificial neural networks (ANN) and logistic regression (LR) models, based on initial clinical parameters, for prediction of mortality after spontaneous ICH. Analysis has been conducted on consecutive dataset of patients with spontaneous ICH, over 5-year period in tertiary care academic hospital. Patients older than 18 years were eligible for inclusion if they had been presented within 6 h from the start of symptoms and had evidence of spontaneous supratentorial ICH on initial brain computed tomography within 24 h. Initial clinical parameters have been used to develop LR and ANN prediction models for hospital mortality as outcome measure. Models have been accessed for discrimination and calibration abilities. We have analyzed 411 patients (199 males and 212 females) with spontaneous ICH, medically treated and not withdrawn from therapy, with average age of 67.35 years. From them, 256 (62.29%) patients died during hospital treatment and 155 (37.71%) patients survived. In the observed dataset, ANN model overall correctly classified outcome in 93.55% of patients, compared with 79.32% of correct classification for the LR model. Discrimination and calibration parameters indicate that both models show an adequate fit of expected and observed values, with superiority of ANN model. Our results favour the ANN model for prediction of mortality after spontaneous ICH. Further studies of the strengths and limitations of this method are needed with larger prospective samples.

[Syringomyelia in demyelinating disease of the central nervous system: report of two cases].

INTRODUCTION: Syringomyelia is a cavitary extension inside the spinal cord which can be either symptomatic or congenitally-idiopathic. Syringomyelia during the course of the disease in patients presenting with clinically definite multiple sclerosis was described earlier. Syringomyelia in patients presenting with a clinically isolated syndrome suggestive of multiple sclerosis is unusual. CASE OUTLINE: We present two patients presenting with demyelinating disease of the central nervous system with syringomyelia in the cervical and thoracic spinal cord. We did not find classical clinical signs of syringomyelia in our patients, but we disclosed syringomyelia incidentally during magnetic resonance exploration. Magnetic resonance exploration using the gadolinium contrast revealed the signs of active demyelinating lesions in the spinal cord in one patient but not in the other. CONCLUSION: Syringomyelia in demyelinating disease of the central nervous system opens the question whether it is a coincidental finding or a part of clinical features of the disease. Differentiation of the significance of syringomyelia finding in these patients plays a role in the choice of treatment concept in such patients.

A case of complex arteriovenous malformation with skull destruction presented in sixth decade of life.

Intracranial AVMs are typically diagnosed before the patient has reached the age of 40 years, and a few cases have been reported of AVM with skull destruction. We described a rare case of a complex cerebral AVM with skull destruction, presented de novo in 52-year-old woman with epileptic seizures. Neuroimaging investigations revealed complex AVM in right hemisphere as well as extracranially, with signs of skull destructions, likely caused by significant involvement of feeders from external carotid artery. Neurosurgery treatment was not recommended due to morphological characteristics and drainage patterns of the AVM. EEG investigation showed discrete specific activity in correspondent area and pharmacology treatment for seizures was initiated. One year after the initial presentation patient had survived rebleeding episode witch left permanent neurology deficit. This patient considered as a rare case of complex AVM with skull destruction, presented de novo in sixth decade of life.

[Influence of low-intensity laser therapy on spatial perception threshold and electroneurographic finding in patients with diabetic polyneuropathy].

INTRODUCTION: Low-intensity laser therapy (LILT) can be applied in cases when patients with diabetic polyneuropathy (DPN) suffer from chronic severe neuropathic pain. OBJECTIVE: We wanted to analyse influence of LILT on spatial perception threshold (SPT) and electroneurographic (ENG) parameters in patients with painful DPN. METHOD: We analysed 45 patients (25 males), average age 54.3 years (54.3 +/- 10.9), with clinical and ENG signs of painful DPN. The patients were divided into two groups: A and B. Group A consisted of 30 patients with DPN who had 30 LILT treatments over the period of 12 weeks and group B consisted of 15 patients with DPN who received only vitamin therapy per os within the same period. Prior to and after 12 weeks of treatment, the following ENG parameters were determined using surface electrodes: motor (MCV) and sensory conduction velocities (SCV) values (in m/s) of nervus (n.) peroneus (NP), n. tibialis (NT) and n. medianus (NM) and their motor distal latency (MDL) values (in ms). SPT value (score as number from 1 to 8) was determined with Tactile Circumferential Discriminator on dorsal part of foot's big toe skin. For statistical analysis, we used Student's t-test and Pearson correlation (sig. 2 tailed) study. RESULTS; We registered statistically significant difference between SPT (p < 0.01) values prior to (5.25 +/- 1.11) and after (4.87 +/- 0.90) LILT, as well as NMMCV (p < 0.05) values prior to (47.18 +/- 5.08) and after (49.12 +/- 3.72) LILT. Besides, we registered, only after LILT, statistically significant correlation beetwen SPT and NMDML (p < 0.01) values and also beetwen SPT and NMSCV (p < 0.05) values. The differences and correlations beetwen other analysed parameters before and after treatments were not significant (p > 0.05). CONCLUSION: In this study we registered significant decrease of SPT and increase of NMMCV after LILT and that indicated a favourable effect of this treatment in analysed patients with painful DPN. In our opinion these results need further investigation.

Sensory-motor index is useful parameter in electroneurographical diagnosis of carpal tunnel syndrome.

It was performed electroneurographic (ENG) studies with surface electrodes and examined nervus medianus (NM) in 60 patients (38 females), average age of 50,28 years (X+/-SD=50,28+/-11), with clinical diagnosis of carpal tunnel syndrome (CTS) and at least one border or discrete abnormal value of conventional electrophysiological tests. It was also examined 57 healthy individuals (33 females) as control group, average age of 45,65 years (X+/-SD=45,65+/-9,68). The sensitivity and specificity of sensory-motor index (SMI), terminal latency index (TLI) and residual latency (RL) were calculated and compared. SMI is determinate by using following formula: distal distance (DD) (in cm)/distal motor latency (DML) (in ms) + sensory conduction velocity (SCV) (in m/s)/motor conduction velocity (MCV) (in m/s) of NM. SCV of NM was measured by antidromic technique in segment wrist-index finger and MCV of NM in forearm segment above wrist. SMI mean value of control group was 3,45 (X+/-SD=3,45+/-0,45) with lower limit of normal value 2,82 and in patients with CTS 2,13 (X+/-SD=2,13 +/-0,37). The sensitivity of SMI in patients with CTS was 98,51%. SMI is useful parameter in electroneurographical diagnosis of CTS and it's determination is easy and fast and specially important in cases with border or discrete abnormal values of other NM electrophysiological parameters, when SMI values can indicate incipient phase of CTS evolution. In rare cases (about 1%) of CTS with selective NM motor axons affection, SMI may have normal value (false negative result), but DML is always prolonged in this cases. SMI is not dependent on age and DD values in patients with CTS and control subjects.