Motivated by recent successes in neural machine translation and image caption generation, we present an end-to-end system to recognize Online Handwritten Mathematical Expressions (OHMEs). Our has three parts: a convolution network for feature extraction, bidirectional LSTM encoding extracted features, attention model generating target LaTex. For recognizing complex structures, our needs large data training. We propose local global distortion models OHMEs from the CROHME database. evaluate on...

Recognition of Handwritten Mathematical Expressions (HMEs) is a challenging problem because the complicated structure and uncommon math symbols contained in HMEs. Moreover, lack training data serious issue, especially for deep learning-based systems. In this paper, we proposed dual loss attention model that utilizes existing latex corpus to improve accuracy. The has two losses, including decoder context matching learn semantic invariant features encoder grammar from handwritten printed MEs....

This paper presents application of deep learning to recognize online handwritten mathematical symbols. Recently various architectures such as Convolution neural network (CNN), Deep (DNN) and Long short term memory (LSTM) RNN have been applied fields computer vision, speech recognition natural language processing where they shown produce state-of-the-art results on tasks. In this paper, we apply max-out-based CNN BLSTM image patterns created from the original patterns, respectively combine...

This paper presents a system for recognizing online handwritten mathematical expressions (MEs) and improvement of structure analysis. We represent MEs in Context Free Grammars (CFGs) employ the Cocke-Younger-Kasami (CYK) algorithm to parse 2D on-line select best interpretation terms symbol segmentation, recognition propose method learn structural relations from training patterns without any heuristic decisions by using two SVM models. stroke order reduce complexity parsing algorithm....

This paper presents deep learning to recognize online handwritten mathematical symbols. Recently various architectures such as Convolution neural networks (CNNs), Deep (DNNs), Recurrent (RNNs) and Long short-term memory (LSTM) RNNs have been applied fields computer vision, speech recognition natural language processing where they shown superior performance state-of-the-art methods on tasks. In this paper, max-out-based CNNs Bidirectional LSTM (BLSTM) are image patterns created from the...

Inspired by recent successes in neural machine translation and image caption generation, we present an attention based encoder decoder model (AED) to recognize Vietnamese Handwritten Text. The composes of three parts: a convolution network (CNN) for extracting invariant features, Bidirectional Long Short-Term Memory (BLSTM) encoding extracted features (BLSTM encoder), (LSTM) with incorporated generating output text (LSTM decoder), which are connected from the CNN part BLSTM finally LSTM...

Recognition of historical documents is a challenging problem due to the noised, damaged characters, and background. However, in Japanese documents, not only contains mentioned problems, pre-modern characters were written cursive are connected. Therefore, character segmentation-based methods do work well. This leads idea creating new recognition system. In this paper, we propose human-inspired document reading system recognize multiple lines documents. During reading, people employ eyes...

Post-processing is an essential step in detecting and correcting errors OCR-generated texts. In this paper, we present automatic OCR post-processing model which comprises both error detection correction phases for output texts of unconstrained Vietnamese handwriting. We propose a hybrid approach generating scoring candidates non-syllable real-syllable based on the linguistic features as well characteristics outputs. evaluate our proposed benchmark database at line level. The experimental...

Different types of OCR errors often occur in texts due to the low quality scanned document images or limitations software. In this paper, we propose a novel unsupervised approach for error correction. Correction candidates are generated and explored their neighborhoods using correction character edits controlled by an adapted hill-climbing algorithm. characters extracted from only original ground truth texts, which do not depend on training data. A weighted objective function used score rank...

본 논문은 N 통신노드들이 다중 안테나 비재생 릴레이국의 협력으로 서로 데이터를 교환하는 다중방향 릴레이 시스템을 제안한다. 기존의 단방향 릴레이를 적용한 통신의 경우 2N 전송 단계가 필요하나 제안하는 시스템은 단일 다중접속 단계와 N-1 방송 단계로 구성된 총 단계만 필요로 한다. 특히 효율적인 단계 전송을 위해 통신노드들을 쌍을 짓고 쌍이 아닌 통신노드들에게 간섭을 주지 않는 새로운 빔형성 행렬과 통신노드 자가간섭 소거기를 평균 합 전송률로 성능을 평가한 결과, 제안 기법이 기존 기법의 향상시키며 그 이득은 노드 수와 수가 비슷할 때 더 커짐을 볼 수 있다. 이러한 기법은 여러 선박간의 정보를 공유할 필요가 있는 해양 통신 환경에서 거리를 확장하는 데 효과적으로 적용할 있을 것으로 기대된다. In this paper, we propose a multi-way relaying system, in which communicating nodes interchange their...

This paper proposes a modified X-Y cut method for reordering strokes of online handwritten mathematical expression (ME) in order to make stroke free recognition. To deal with overlapping, which causes problems the method, we determine vertically ordered by detecting vertical symbols and its upper/lower MEs. An upper ME lower are treated as MEs reordered recursively. Unordered on left side symbol horizontally strokes. The remaining from right top bottom. results evaluations CROHME 2014...

This paper presents a semi-incremental recognition method for online handwritten mathematical expressions (MEs). The reduces the waiting time after an ME is written until result of output. Our has two main processes, one to process latest stroke, other find and correct wrong recognitions in strokes up stroke. In first process, segmentation, Cocke-Younger-Kasami (CYK) algorithm are only executed second all previous segmentations updated if they significantly changed stroke input, then,...

This paper proposes a method for speeding upparsing process recognizing online handwritten mathematicalexpressions (OHME). We prune infeasible partitions in theparsing table to reduce the time parsing process. Lowscore are candidates pruning. Our can beapplied any algorithms that use score functions. Inthis paper, we stroke order free system as baseline system.The is follows. First, analyze scores ofpartitions each row of table. Then, determine athreshold low partitions. Finally, weemploy...

OCR post-processing is an important step for improving the quality of output texts. Long short-term memory (LSTM) a deep learning model, which has wide-range applications in many domains like time series prediction, natural language processing and speech recognition. In this paper, we propose error correction model using neural machine translation with bidirectional LSTM networks at syllable level. Vietnamese text dataset evaluation outputted from engine based on attention-based...