A Bayesian model for recognizing handwritten mathematical expressions

MacLean, Scott; Labahn, George

doi:10.1016/j.patcog.2015.02.017

Cited by 32 publications

(6 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Methods of HMER Traditional HMER methods require specially designed grammatical rules to represent the twodimensional structural information of formulas, such as graph grammar (Lavirotte and Pottier 1998), attributive clause grammar (Chan and Yeung 2001), relational grammar (MacLean and Labahn 2013) or probabilistic model based grammar (Awal, Mouchere, and Viard-Gaudin 2014;MacLean and Labahn 2015;Álvaro, Sánchez, and Benedí 2016).…”

Section: Related Workmentioning

confidence: 99%

Handwritten Mathematical Expression Recognition via Attention Aggregation Based Bi-directional Mutual Learning

Bian

Qin

Xin

et al. 2022

AAAI

View full text Add to dashboard Cite

Handwritten mathematical expression recognition aims to automatically generate LaTeX sequences from given images. Currently, attention-based encoder-decoder models are widely used in this task. They typically generate target sequences in a left-to-right (L2R) manner, leaving the right-to-left (R2L) contexts unexploited. In this paper, we propose an Attention aggregation based Bi-directional Mutual learning Network (ABM) which consists of one shared encoder and two parallel inverse decoders (L2R and R2L). The two decoders are enhanced via mutual distillation, which involves one-to-one knowledge transfer at each training step, making full use of the complementary information from two inverse directions. Moreover, in order to deal with mathematical symbols in diverse scales, an Attention Aggregation Module (AAM) is proposed to effectively integrate multi-scale coverage attentions. Notably, in the inference phase, given that the model already learns knowledge from two inverse directions, we only use the L2R branch for inference, keeping the original parameter size and inference speed. Extensive experiments demonstrate that our proposed approach achieves the recognition accuracy of 56.85 % on CROHME 2014, 52.92 % on CROHME 2016, and 53.96 % on CROHME 2019 without data augmentation and model ensembling, substantially outperforming the state-of-the-art methods. The source code is available in https://github.com/XH-B/ABM.

show abstract

Section: Related Workmentioning

confidence: 99%

Handwritten Mathematical Expression Recognition via Attention Aggregation Based Bi-directional Mutual Learning

Bian

Qin

Xin

et al. 2022

AAAI

View full text Add to dashboard Cite

show abstract

“…To present grouping and bunching systems and execution investigation with exactness in blunder detection.Spatial parallel order, Spatial summed up direct model (SGLM) [2] and the Bayesian spatial summed up straight blended model (SGLMM) is utilized to recoup information robustness. A Bayesian characterization for perceiving written by hand numerical articulations [3]. Presenting some constraintson how data sources might be parceled, [4]we inferred an effective parsing calculation got from Unger's method.Expert elicitation and Bayesian System demonstrating for transportation Mishaps.…”

Section: Literature Surveymentioning

confidence: 99%

A Supervised Classification Techniques to Optimize Error Evaluation and Space Complexity

2019

IJITEE

View full text Add to dashboard Cite

Bayesian classification is based on Baye’s Theorem, which is applied on a conditional probability basis of posterior and prior probabilities in parallel with future evidence. Prior Probabilities are the original probabilities of an outcome which will be updated with new information to create posterior probability. The revised probability of an event occurring after taking into consideration new information.A Bayesian classifier is used to predict the values of features for members of that class. It is used to over come the diagnostic and predictive problems. This classification provides a useful perspective for understanding and evaluating machine learning algorithms.It is a probabilistic learning algorithm which calculates the explicit probabilities for hypothesis, among the most common learning problem.The proposed work has focused on designing of two classification algorithms naïve space and naïve Mine classification to optimize space complexity and error evaluation forlarger data sets.

show abstract

“…The recognition of a word involves the pattern, pose, matching, identification, or object discrimination. Nevertheless, an efficient method of dealing with recognition is through an efficient representation model [17].…”

Section: Exploits Maximally Stable Extremal Regions (Msers) Which Promentioning

confidence: 99%

Parallel Processing Based on Fragmentation of English Text Pictures to Speed Up Recognition of Characters on a Multi-Core Computer

Ali¹,

Rahma²

2019

djps

View full text Add to dashboard Cite

Script picture analyses use algorithms to turn images of text into readable computer data. The goal of script picture analyses is pattern recognition of script and graphics combinations from images for extraction of the data. This paper solves a long-existing problem in pattern recognition in script images regarding recognition of characters and words using a computer processor. The proposed system rely on main steps of execute, first step: is load English script of image file, second step: is partition image file to several image pages, third step: is divide the image into number of blocks, and final step: multi-block method can be used in each row of the script image with parallel processing using a multi-core computer (one processor P1, two processors P2, three processors P3, and four processors P4) with rapid processing for script images for pattern recognition. The outcome obtained with a multi-core computer is very fast for processing script images and is efficient and accurate. It can use the modified algorithm in multi-processor system that available now and not used all the capacity in processing in one processor as in old computers.

show abstract

A Bayesian model for recognizing handwritten mathematical expressions

Cited by 32 publications

References 15 publications

Handwritten Mathematical Expression Recognition via Attention Aggregation Based Bi-directional Mutual Learning

Handwritten Mathematical Expression Recognition via Attention Aggregation Based Bi-directional Mutual Learning

A Supervised Classification Techniques to Optimize Error Evaluation and Space Complexity

Parallel Processing Based on Fragmentation of English Text Pictures to Speed Up Recognition of Characters on a Multi-Core Computer

Contact Info

Product

Resources

About