GitPlanet

Cheap and reliable Node.js hosting starts at $3/month, and $1/month static HTML hosting

Created with love in Canada, visit hostnodejs.com today

Feel like to post an Ad? Learn Details
All Projects → vuptran → deep-spell-checkr

vuptran / deep-spell-checkr

Licence: MIT license
Keras implementation of character-level sequence-to-sequence learning for spelling correction

Programming Languages

python
139335 projects - #7 most used programming language

Labels

sequence-to-sequencespelling-correctionlstm-neural-networks

Projects that are alternatives of or similar to deep-spell-checkr

Sequence-to-Sequence-Learning-of-Financial-Time-Series-in-Algorithmic-Trading
My bachelor's thesis—analyzing the application of LSTM-based RNNs on financial markets. 🤓
Stars: ✭ 64 (-1.54%)
Mutual labels:  sequence-to-sequence, lstm-neural-networks
Word-Level-Eng-Mar-NMT
Translating English sentences to Marathi using Neural Machine Translation
Stars: ✭ 37 (-43.08%)
Mutual labels:  sequence-to-sequence, lstm-neural-networks
sheldon
Very Simple Erlang Spell Checker
Stars: ✭ 63 (-3.08%)
Mutual labels:  spelling-correction
SSIM Seq2Seq
SSIM - A Deep Learning Approach for Recovering Missing Time Series Sensor Data
Stars: ✭ 32 (-50.77%)
Mutual labels:  sequence-to-sequence
privapi
Detect Sensitive REST API communication using Deep Neural Networks
Stars: ✭ 42 (-35.38%)
Mutual labels:  lstm-neural-networks
ForestCoverChange
Detecting and Predicting Forest Cover Change in Pakistani Areas Using Remote Sensing Imagery
Stars: ✭ 23 (-64.62%)
Mutual labels:  sequence-to-sequence
cross-lingual-open-ie
MT/IE: Cross-lingual Open Information Extraction with Neural Sequence-to-Sequence Models
Stars: ✭ 22 (-66.15%)
Mutual labels:  sequence-to-sequence
lstm-kalman-hybrid-timeseries
Hybrid Time Series using LSTM and Kalman Filtering
Stars: ✭ 33 (-49.23%)
Mutual labels:  lstm-neural-networks
parse seq2seq
A tensorflow implementation of neural sequence-to-sequence parser for converting natural language queries to logical form.
Stars: ✭ 26 (-60%)
Mutual labels:  sequence-to-sequence
Keras-LSTM-Trajectory-Prediction
A Keras multi-input multi-output LSTM-based RNN for object trajectory forecasting
Stars: ✭ 88 (+35.38%)
Mutual labels:  sequence-to-sequence
tesla-stocks-prediction
The implementation of LSTM in TensorFlow used for the stock prediction.
Stars: ✭ 51 (-21.54%)
Mutual labels:  lstm-neural-networks
DeepLog
This is the realization of core DeepLog
Stars: ✭ 29 (-55.38%)
Mutual labels:  lstm-neural-networks
recurrent-neural-net
A recurrent (LSTM) neural network in C
Stars: ✭ 68 (+4.62%)
Mutual labels:  lstm-neural-networks
spell
Spelling correction and string segmentation written in Go
Stars: ✭ 24 (-63.08%)
Mutual labels:  spelling-correction
parapred
Paratope Prediction using Deep Learning
Stars: ✭ 49 (-24.62%)
Mutual labels:  lstm-neural-networks
ocr-machine-learning
OCR Machine Learning in python
Stars: ✭ 42 (-35.38%)
Mutual labels:  spelling-correction
malware api class
Malware dataset for security researchers, data scientists. Public malware dataset generated by Cuckoo Sandbox based on Windows OS API calls analysis for cyber security researchers
Stars: ✭ 134 (+106.15%)
Mutual labels:  lstm-neural-networks
ismir2019-music-style-translation
The code for the ISMIR 2019 paper “Supervised symbolic music style translation using synthetic data”.
Stars: ✭ 27 (-58.46%)
Mutual labels:  sequence-to-sequence
LSTM-footballMatchWinner
This repository contains the code for a conference paper "Predicting the football match winner using LSTM model of Recurrent Neural Networks" that we wrote
Stars: ✭ 44 (-32.31%)
Mutual labels:  lstm-neural-networks
WordSegmentationDP
Word Segmentation with Dynamic Programming
Stars: ✭ 18 (-72.31%)
Mutual labels:  spelling-correction
View All Similar Projects ➔

Sequence-to-Sequence Learning for Spelling Correction

This repository contains a Keras implementation of an encoder-decoder LSTM architecture for sequence-to-sequence spelling correction. The character-level spell checker is trained on unigram tokens derived from a vocabulary of more than 466k 33k unique English words. After 12 one hour of training, the speller achieves an accuracy performance of 96.7% 97.6% on a validation set comprised of more than 26k tokens.

Input sentence:
> The rabbit holV ewnt straight on liek a tnnel ofr some way any then dipped suddnely down so suddnenly tat Alice had nobt a moment to think aPout stopipng herself before she found hersefl falling dow a verZy deeup wLell

Decoded sentence:
> The rabbit hole went straight on like a tunnel for some way any then dipped suddenly drown so suddenly tat Alice had nob a moment to think Pout stopping herself before she found herself falling down a very deep well

Target sentence:
> The rabbit hole went straight on like a tunnel for some way and then dipped suddenly down so suddenly that Alice had not a moment to think about stopping herself before she found herself falling down a very deep well

Some ideas for further improvement on speller performance:

  • Implement an attention mechanism
  • Incorporate beam search as part of the loss evaluation
  • Extend to high-order ngrams to capture intra-word contexts

Requirements

The code is tested on Ubuntu 16.04 with the following components:

Software

  • Anaconda Python 3.6
  • Keras 2.2.4 using TensorFlow GPU 1.12.0 backend
  • CUDA 9.1 with CuDNN 7.1

Optional Hardware

  • Intel Xeon CPU with 32 cores
  • 64GB of system RAM
  • NVIDIA GeForce GTX TITAN X GPU with 12GB of VRAM

Acknowledgment

The idea behind this project is inspired by this blog post, with several distinct enhancements in this implementation, such as teacher forcing, that result in much faster convergence and better performance.

Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].