As the incidence of this disease has increased significantly in recent years, expert systems and machine learning techniques to problem have also taken a great attention from many scholars. This study aims at diagnosing prognosticating breast cancer with method based on random forest classifier feature selection technique. By weighting, keeping useful features removing redundant datasets, was obtained solve diagnosis problems via classifying Wisconsin Breast Cancer Diagnosis Dataset...

This paper presents an improvement in recognizing offline handwritten mathematical expressions (HMEs) by deep neural networks. We train it end-to-end using weakly supervised learning. The network has three parts: encoder a Convolutional Neural Network to encode high-level features from input HME image; decoder gated recurrent units with attention parse the and generate output expression LaTeX format; symbol classifier improve localization classification of features. Besides, we use model...

This paper presents methods for three different tasks of recognizing anomalously deformed Kana in Japanese historical documents, which were contested by IEICE PRMU1 2017. The have levels: single character recognition, characters sequence recognition and unrestricted recognition. We compare several each task. For the level 1, we evaluate CNN based BLSTM methods. 2, consider variations a combined architecture BLSTM. 3, an extension method 2 segmentation method. achieve accuracy 96.8%, 87.12%...

This paper presents a model of Deep Convolutional Recurrent Network (DCRN) for recognizing offline handwritten Japanese text lines without explicit segmentation characters. Most traditional recognizers perform image into characters before individually each character. Although by recognition and context are employed to recover from errors, errors made at this stage directly make an impact on the performance whole system. The DCRN consists three parts: convolutional feature extractor using...

This paper presents an end-to-end model of Deep Convolutional Recurrent Network (DCRN) for recognizing offline handwritten Japanese text lines. The DCRN has three parts: a convolutional feature extractor using Neural (DCNN) to extract sequence from line image; recurrent layers employing Bidirectional LSTM predict pre-frame the sequence; and transcription layer Connectionist Temporal Classification (CTC) convert predictions into label sequence. Since our requires large data training, we...

This paper presents the results of VOHTR 2018 competition on Vietnamese Online Handwritten Text Recognition. The goal this is to evaluate and compare recent online handwritten text recognition systems which contains many delayed strokes caused by diacritic marks. Besides, general objective encourage studies based large handwriting database collected from 200 writers. In competition, we introduce three tasks consisting word (task 1), text-line 2) paragraph 3) are described in details....

This paper presents an attention-based convolutional sequence to (ACseq2seq) model for recognizing input image of multiple text lines from Japanese historical documents without explicit segmentation lines. The recognition system has three main parts: a feature extractor using Convolutional Neural Network (CNN) extract image; encoder employing bidirectional Long Short-Term Memory (BLSTM) encode the sequence; and decoder unidirectional LSTM with attention mechanism generate final target based...

We propose a deep neural network-based method to recover dynamic online trajectories from offline handwritten Japanese kanji character images. It is challenging task since characters consist of multiple strokes. Our proposed model has three main components: Convolutional Neural Network-based encoder, Long Short-Term Memory decoder with an attention layer, and Gaussian Mixture Model (GMM). The encoder focuses on feature extraction while the refers extracted features generates time-sequences...

Despite of recent breakthroughs in the accuracy single character recognition using deeper convolution neural networks, one remaining problems is that OCRs almost fail to recognize patterns when they are severely degraded, especially those historical documents. Another problem characters documents lack sufficient training because heavy cost for annotation. This paper proposes a attention generative adversarial network named CAGAN restoring heavily degraded so improve their and even help...

This paper presents a semi-incremental recognition method for online handwritten English text. We employ local processing strategy and focus on recent sequence of strokes defined as "scope". For the latest scope, we build update segmentation candidate lattice advance best-path search incrementally. utilize result in previous scope to exclude unnecessary candidates. reduces number word with reduced time. also reuse scope. Moreover, triggering processes every few save CPU Experiment made...

This paper presents an incremental recognition method for online handwritten mathematical expressions (MEs), which is used busy interface (recognition while writing) without large waiting time. We employ local processing strategy and focus on recent strokes. For the latest stroke, we perform segmentation, update of Cocke-Younger-Kasami (CYK) table. also reuse segmentation candidates in previous processes. Moreover, using multi-thread reduces Experiments our data set show effectiveness not...

Discriminating ambiguous symbols in online handwritten mathematical expression is difficult without context. We propose a Bidirectional Recurrent Neural Network for segmenting and classifying symbols. The context from forward backward directions helps the classification model discriminate improve recognition rates. integrated into Stochastic Context-Free Grammar system recognizing expressions. show effectiveness of approach improving symbol segmentation on CROHME 2016 dataset.