A neural fuzzy training approach for improving speech recognition

Eight horses with previous racing experience were used in a comparative study of training methods for Thoroughbred racehorses. They were randomly assigned to two groups of four horses each. One group was trained using an interval training method (IT) and the other using conventional training (CT) methods. Peak heart rates, heart rate recovery curves, peak plasma lactate levels, plasma lactate clearance rates and run times were used to evaluate differences in the training methods. Peak heart rates, heart rate recovery curves, and run times were not significantly different between the groups. However, higher lactate production and increased plasma lactate clearance by the IT group demonstrated an increased anaerobic capacity.

This paper describes a new training method for a continuous mapping and/or pattern classification neural network that performs local sample-density smoothing. A conventional training method uses point-to-point mapping from an input space to an output space. Even though the mapping may be precise at two given training sample points, there are no guarantees of mapping accuracy at points on a line segment connecting the sample points. This paper first discusses a theory for formulating line-to-line mapping. The theory is called interpolation training. This paper then expands the theory to k-nearest neighbor interpolation. The k-neighbor interpolation training (KNIT) method connects an input sample training point to its k-neighbor points via k line segments. Then, the method maps these k line segments in the input space for each training sample to linear line segments in the output space that interpolate between training output values. Thus, a web structure made by connecting input samples is mapped into the same structure in an output space. The KNIT method reduces the over learning problem caused by point-to-point training by smoothing input/output functions. Simulation tasks show that KNIT improves vowel recognition on a small speech database.

Neural network (NN) based prediction or detection systems often perform excellently with easy problems without considering 1-5% difficult problems. This work proposes an adversarial NN training method for constructing the prediction interval (PI). The proposed training method considers adverse situations where traditional NN based PIs frequently fail. First, the conventional lower upper bound estimation (LUBE) method is applied in parallel for initial training of NNs with different initialization. Each NN based PI fails to cover a few samples. Input combinations of those samples are adversely changed by a small amount to generate the adverse samples. A new dataset is generated by appending adverse samples. Finally, an NN is trained with the adverse dataset. The method is applied to construct the NN for wind power prediction. According to the result analysis, the proposed method performs better in adverse situations.

The Polynomial Cellular Neural Network (PCNN) is a fully parallel, scalable, non-linear processor that uses polynomial terms to solve non-linear problems in a lattice fashion. The parallel nature of such processor allows every neuron (or cell) to gather information from the nearby neurons and independently process the retrieved values by employing non-linear functions and synaptic weights. Nonetheless, one of the main challenges of the PCNN is the determination of the synaptic weights in order to achieve the desired behavior. In this paper, a new training method is presented, based on two fundamental concepts: the root location training method and the polynomial surfaces. The proposed training method is able to straightforwardly determine the requested synaptic weights for any outer-totallistic cellular automata behavior. In order to deliver a proof of the potential of such proposition, several image processing tasks are performed with a single layered PCNN.

This work addresses the task of action recognition in video sequences. In real world applications, this task is quite challenging due to the complex background of video content, the similarities between different types of actions, the dependence on a large amount of annotated data, and so on. Most of the existing methods fail to distinguish similar actions with the same static appearance and motion pattern. We attempt to address this issue from the perspective of a local-global view, considering videos as combinations of a set of action units (local semantic information) and their relations along temporal dimension (global relation information). To achieve this end, we propose a novel Local-global Networks (LgNet) to enhance recognition of similar action. Besides, we propose an end-to-end training method to decrease the reliance on annotated data. It combines self-supervised learning and supervised learning, which not only enables the model to learn video representations from a large number unannotated data but also avoids subsequent finetuning. The proposed training method can be flexibly equipped to a wide array of vision tasks. Experiments on several benchmark datasets show that our proposed model and training method achieve state-of-the-art performance.

Large imaging surveys will rely on photometric redshifts (photo-z's), which are typically estimated through machine-learning methods. Currently planned spectroscopic surveys will not be deep enough to produce a representative training sample for Legacy Survey of Space and Time (LSST), so we seek methods to improve the photo-z estimates that arise from nonrepresentative training samples. Spectroscopic training samples for photo-z's are biased toward redder, brighter galaxies, which also tend to be at lower redshift than the typical galaxy observed by LSST, leading to poor photo-z estimates with outlier fractions nearly 4 times larger than for a representative training sample. In this Letter, we apply the concept of training sample augmentation, where we augment simulated nonrepresentative training samples with simulated galaxies possessing otherwise unrepresented features. When we select simulated galaxies with (g-z) color, i-band magnitude, and redshift outside the range of the original training sample, we are able to reduce the outlier fraction of the photo-z estimates for simulated LSST data by nearly 50% and the normalized median absolute deviation (NMAD) by 56%. When compared to a fully representative training sample, augmentation can recover nearly 70% of the degradation in the outlier fraction and 80% of the degradation in NMAD. Training sample augmentation is a simple and effective way to improve training samples for photo-z's without requiring additional spectroscopic samples.

In this study, performances comparison to discriminate five mental states of five artificial neural network (ANN) training methods were investigated. Wavelet Packet Transform (WPT) was used for feature extraction of the relevant frequency bands from raw electroencephalogram (EEG) signals. The five ANN training methods used were (a) Gradient Descent Back Propagation (b) Levenberg-Marquardt (c) Resilient Back Propagation (d) Conjugate Learning Gradient Back Propagation and (e) Gradient Descent Back Propagation with movementum.

Within the field of seismic data acquisition with active sources, the technique of acquiring simultaneous data, also known as blended data, offers operational advantages. The preferred processing of blended data starts with a step of deblending, that is separation of the data acquired by the different sources, to produce data that mimic data from a conventional seismic acquisition and can be effectively processed by standard methods. Recently, deep learning methods based on the deep neural network have been applied to the deblending task with promising results, in particular using an iterative approach. We propose an enhancement to deblending with an iterative deep neural network, whereby we modify the training stage of the deep neural network in order to achieve better performance through the iterations. We refer to the method that only uses the blended data as the input data as the general training method. Our new multi‐data training method allows the deep neural network to be trained by the data set with the input patches composed of blended data, noisy data with low amplitude crosstalk noise, and unblended data, which can improve the ability of the deep neural network to remove crosstalk noise and protect weak signal. Based on such an extended training data set, the multi‐data training method embedded in the iterative separation framework can result in different outputs at different iterations and converge to the best result in a shorter iteration number. Transfer learning can further improve the generalization and separation efficacy of our proposed method to deblend the simultaneous‐source data. Our proposed method is tested on two synthetic data and two field data to prove the effectiveness and superiority in the deblending of the simultaneous‐source data compared with the general training method, generic noise attenuation network and low‐rank matrix factorization methods.

Background and Aim: With the continuous development of the project and the increasing confrontation and competitiveness of this competition, the physical fitness of the players is being given higher requirements. Therefore, the objective of this research was to compare it with the traditional training method. The combination of digital technology and traditional table tennis strength, agility, and quickness training methods are more effective. Materials and Methods: The population used in this research was 120 students at Jing Zhou City Sports School. According to the formula of Yamane, the total was 92 samples. This research introduces the training programs for table tennis players, focusing on the theories of training strength, agility, and quickness. Semi-structured interviews were conducted to determine the training plan, and the final indicators were obtained through a questionnaire survey. The training experiment lasted for 8 weeks. Results: The research results show that: (1) Compared with the conventional lower limb mobility training, the compound training method is better in sports load monitoring. (2) Compared with the conventional lower limb mobility training method, table tennis coaches and athletes pay more attention to the integration of digital technology compound training methods in improving the ability of lower limb mobility. Conclusion: The research findings were: (1) the Effect of compound training on half-meter word test performance, (2) the Effect of compound training on the performance of side slip plus linear sprint test, (3) the Effect of compound training on SJ squat jump test performance.

We present a novel large-context end-to-end automatic speech recognition (E2E-ASR) model and its effective training method based on knowledge distillation. Common E2E-ASR models have mainly focused on utterance-level processing in which each utterance is independently transcribed. On the other hand, large-context E2E-ASR models, which take into account long-range sequential contexts beyond utterance boundaries, well handle a sequence of utterances such as discourses and conversations. However, the transformer architecture, which has recently achieved state-of-the-art ASR performance among utterance-level ASR systems, has not yet been introduced into the large-context ASR systems. We can expect that the transformer architecture can be leveraged for effectively capturing not only input speech contexts but also long-range sequential contexts beyond utterance boundaries. Therefore, this paper proposes a hierarchical transformer-based large-context E2E-ASR model that combines the transformer architecture with hierarchical encoder-decoder based large-context modeling. In addition, in order to enable the proposed model to use long-range sequential contexts, we also propose a large-context knowledge distillation that distills the knowledge from a pre-trained large-context language model in the training phase. We evaluate the effectiveness of the proposed model and proposed training method on Japanese discourse ASR tasks.

In the domain of artificial neural networks, it is important to know what their representation, classification and generalization capabilities are. There is also a need for time and resource-efficient training algorithms. Here, a new zero-error training method is derived for digital computers and single hidden layer networks. This method is the least upper bound on the number of hidden neurons as well. The bound states that if there are N input vectors expressed as rational numbers, a network having N − 1 neurons in the hidden layer and M neurons at the output represents a bounded function F: RD→RM for all input vectors. Such a network has massively shared weights calculated by 1 + M regular systems of linear equations. Compared to similar approaches, this new method achieves a theoretical least upper bound, is fast, robust, adapted to floating-point data, and uses few free parameters. This is documented by theoretical analyses and comparative tests. In theory, this method provides a new constructional proof of the least upper bound on the number of hidden neurons, extends the classes of supported activation functions, and relaxes conditions for mapping functions. Practically, it is a non-iterative zero-error training algorithm providing a minimum number of neurons and layers.

Aiming at the problems of traditional sparse representation algorithms that has a large amount of calculation and low recognition rate under the condition of small samples, a sparse representation classification algorithm (PCA-SRC) combined with principal component feature extraction is proposed. The algorithm can maintain the manifold structure of the finger vein space, solve the singularity problem of the characteristic equation matrix in small sample image recognition. Principal component analysis (PCA) is used to reduce the dimension of the feature matrix to eliminate redundancy. Construct an over-complete dictionary of sparse representation, and design a classifier based on sparse representation to get the classification results. The finger vein feature matching experiment is carried out using the self-built finger vein data set, and the algorithm proposed in this paper is compared with the traditional algorithm. The results show that the recognition rate of the four algorithms can reach more than 94%. The fusion of PCA and SRC speeds up the recognition speed. The recognition rate is higher than the traditional algorithm under different numbers of training samples, and the best recognition rate is 100%. The algorithm has high finger vein recognition rate and low dependence on the training sample set, which provides a new idea for the realization of finger vein recognition.

The sparse representation can deal with the lack of sample problem due to utilizing of all the training samples. However, the discrimination ability will degrade when more training samples are used for representation. We propose a novel appearance-based palmprint recognition method. We aim to find a compromise between the discrimination ability and the lack of sample problem so as to obtain a proper representation scheme. Under the assumption that the test sample can be well represented by a linear combination of a certain number of training samples, we first select the representative training samples according to the contributions of the samples. Then we further refine the training samples by an iteration procedure, excluding the training sample with the least contribution to the test sample for each time. Experiments on PolyU multispectral palmprint database and two-dimensional and three-dimensional palmprint database show that the proposed method outperforms the conventional appearance-based palmprint recognition methods. Moreover, we also explore and find out the principle of the usage for the key parameters in the proposed algorithm, which facilitates to obtain high-recognition accuracy.

Nambya language has been scantly researched on. While it has been introduced at primary, secondary and tertiary level, there is no consensus on the agreed phonemes. With the background of learning a Language of Wider Communication (LWC), the LWC has influenced the way Nambya is written or spelt. This paper investigates the phonemes in Nambya, a previously marginalised language spoken by an estimated 117 000 inhabitants (https://joshuaproject.net/people_groups / -retrieved 16 June 2022.) found in Hwange district. The study employed a qualitative design and was conducted in Hwange district in western province of Zimbabwe. The study attempted a systematic description and classification of Nambya consonants, and vowels using a descriptive framework. This research aimed to establish and provide a detailed study of the consonant sounds available in the language. The target population were the mother tongue speakers who had either learnt the language at formal setups or were natural speakers of the language. Expert and homogeneous purposive sampling techniques were applied to select participants and the sample size was ten. Data was generated through semi structured interviews and focused group discussions. It was also collected through primary and secondary sources and analysed according to consonantal and vowel structures of the language. The study found that Nambya has basic 5 vowel system. The consonant structure includes single and modified consonants. Long vowels are restricted to (i) the future tense, future (Incidental) Indicative, affirmative, other persons, (ii)in the third person singular of negative tenses and present indicative negative.

Logarithmic barrier methods are developed for training supervised neural networks. Two algorithms are presented: a deterministic logarithmic barrier and a stochastic logarithmic barrier. Specifically, we consider neural network training as a nonlinear programming problem and use barrier methods to find the optimal weights. Furthermore, we put constraints on the weights to avoid network paralysis. The search direction is derived using a recursive prediction error method (RPEM) that approximates the inverse of the Hessian of a logarithmic error function iteratively. The weights move on a center trajectory in the interior of the feasible weight space and have good convergence properties. In the case of the stochastic version, random fluctuations are added to the weights in order to escape local minima. Ill-conditioning problems are discussed and computational experiments are provided.Keywordsneural networksdeterministic and stochastic logarithmic barrier algorithmsinterior point methodsNewton type methodsneural network trainingANN training parallelization