Increasingly, pattern recognition is being employed in expansive information systems. The synergy of progress in image processing theory and access to open-source libraries enables the implementation of creative solutions for diverse practical challenges. One such challenge includes the automated processing of responses from large-scale examinations. This paper was considered to be focused on performance improvement methods of form recognition systems by using parallel processing techniques.

The utilization of pattern recognition is on the rise extensively in information systems. The convergence of progress in image processing and the accessibility of open-source libraries enables the implementation of innovative solutions for diverse practical problems. One notable challenge pertains to automatically processing responses in mass large-scale exams. This paper introduces a developed system tailored for recognizing such exam results, showcasing its capacity to deliver dependable, effective, and impartial assessments. This system can be configured on almost any type of form. Its use also allows you to abandon the expensive and difficult to use OMR scanners. To increase productivity of system we propose to use the multicore/multithreading property of modern processors to parallelize processes within a single workstation. As a result of experiments, itwas found that the transition to multi-threaded recognition can increase productivity up to 3.5 times in comparison with single-threaded. To reduce the physical size of exam cards, it is proposed to fill in the answers with handwritten symbols instead of filling in the circles. Multilayer and convolutional neural networks were used as a recognition module. A comparative evaluation of the dependence of recognition results on the architecture of neural networks and the feature extraction algorithm was carried out.

Pattern recognition is becoming increasingly utilized within extensive information systems. The combination of advancements in image processing theory and the availability of open source libraries allows for the application of innovative solutions to various practical problems. One such problem involves automatically processing the answers from large-scale exams. This paper presents a developed system designed specifically for handling the results of these exams, demonstrating its ability to provide reliable, efficient, and unbiased assessments.

Introduction. The implementation of information technologies in various spheres of public life dictates the creation of efficient and productive systems for entering information into computer systems. In such systems it is important to build an effective recognition module. At the moment, the most effective method for solving this problem is the use of artificial multilayer neural and convolutional networks. The purpose of the paper. This paper is devoted to a comparative analysis of the recognition results of handwritten characters of the Azerbaijani alphabet using neural and convolutional neural networks. Results. The analysis of the dependence of the recognition results on the following parameters is carried out: the architecture of neural networks, the size of the training base, the choice of the subsampling algorithm, the use of the feature extraction algorithm. To increase the training sample, the image augmentation technique was used. Based on the real base of 14000 characters, the bases of 28000, 42000 and 72000 characters were formed. The description of the feature extraction algorithm is given.Conclusions. Analysis of recognition results on the test sample showed:as expected, convolutional neural networks showed higher results than multilayer neural networks;the classical convolutional network LeNet-5 showed the highest results among all types of neural networks. However, the multi-layer 3-layer network, which was input by the feature extraction results; showed rather high results comparable with convolutional networks;there is no definite advantage in the choice of the method in the subsampling layer. The choice of the subsampling method (max-pooling or average-pooling) for a particular model can be selected experimentally;increasing the training database for this task did not give a tangible improvement in recognition results for convolutional networks and networks with preliminary feature extraction. However, for networks learning without feature extraction, an increase in the size of the database led to a noticeable improvement in performance. Keywords: neural networks, feature extraction, OCR.

In this work, we applied Adaptive Neuro-Fuzzy Inference System to three different classification problems: (1) sentence-level subjectivity detection, (2) sentiment analysis of texts, and (3) detecting user intention in natural language call routing system. We used English dataset for the first and second problems, but Azerbaijani dataset for the third problem based on same features. Our feature extraction algorithm calculates a feature vector based on the statistical occurrences of words in a corpus without any lexical knowledge.

The context analysis of customer requests in a natural language call routing problem is investigated in the paper. One of the most significant problems in natural language call routing is a comprehension of client request. With the aim of finding a solution to this issue, the Hybrid HMM and ANFIS models become a subject to an examination. Combining different types of models (ANFIS and HMM) can prevent misunderstanding by the system for identification of user intention in dialogue system. Based on these models, the hybrid system may be employed in various language and call routing domains due to non-usage of lexical or syntactic analysis in classification process.

The article is devoted to the investigation of speech recognition in Flight Simulator cockpit. We have done research and developed software for speech recognition in Flight Simulator with limited vocabulary in C# from scratch. The speech recognition system works in real time and offline mode in Windows platform. We used Mel Frequency Cepstral Coefficients and Linear Predictive Coefficients feature extraction algorithms and trained the system by multilayer artificial neural networks. The User Interface of platform is highly functional and allows users to update system parameters through the interface. Our Speech Recognition system results are compared to the results of Microsoft Speech SDK and received satisfactory achievement.

In the work, the problem of training of deep neural networks on GPU and CPU-based frameworks is investigated. As a test problem, the MNIST Dataset was taken for recognition of handwritten digits. The results of deep learning are compared and analyzed in both frameworks for the current problem.

In previous work [1], it was shown that NeuroFuzzy Models (ANFIS) applied to dialog analysis could determine user intent with reasonable accuracy. Hidden Markov models (HMMs) achieved comparable accuracy, but with a different pattern of errors. A hybrid approach that fused the two methods was more accurate that either alone. This technique has been modified to extract sentiment from the "Rotten Tomatoes" movie review database. The reported systems include HMM only, ANFIS only, and a hybrid of the two. The two single-component systems each perform 82-83% correct results from unedited reviews. The hybrid system is able to improve accuracy by a full percentage point, achieving 84% correct. It is anticipated that when an automatic editing module is inserted, accuracy will improve to a level commensurate with human judgment.

Hidden Markov models are a powerful statistical tool and have been used in many areas of speech and natural language processing. In this work, we attempt to detect sentence-level subjectivity by means of hidden Markov model which hasn't been thoroughly investigated for subjectivity analysis. Our feature extraction algorithm calculates a feature vector based on the statistical occurrences of words in a corpus without any linguistic knowledge except tokenization. For this reason, this model can be applied to any language; i.e., there is no lexical, grammatical, syntactical analysis used in the classification process.

In this work, we attempt to detect sentencelevel subjectivity by means of two supervised machine learning approaches: a Fuzzy Control System and Adaptive Neuro-Fuzzy Inference System. Even though these methods are popular in pattern recognition, they have not been thoroughly investigated for subjectivity analysis. We present a novel “Pruned ICF Weighting Coefficient,” which improves the accuracy for subjectivity detection. Our feature extraction algorithm calculates a feature vector based on the statistical occurrences of words in a corpus without any lexical knowledge. For this reason, these machine learning models can be applied to any language; i.e., there is no lexical, grammatical, syntactical analysis used in the classification process.