TAGM

KO | EN

gitlite — search

#hacktoberfest #ai #typescript #llm #go #android #machine-learning #rust #linux #docker #macos #windows #awesome-list #pytorch

(★ 21)

The source code for Temporal Attention-Gated Model.

TAGM Latest Version Download

Download Latest Version (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

This is the implementation of the Temporal Attention-Gated Model, which is proposed in the following paper:
Wenjie Pei, Tadas Baltrušaitis, David M.J. Tax and Louis-Philippe Morency. "Temporal Attention-Gated Model for Robust Sequence Classification", https://arxiv.org/pdf/1612.00385.pdf, which is accepted by CVPR 2017.

The code is implemented in Lua and Torch. It contains mainly the following parts:   
* main.lua:   the starting point of the entire code. 
* train_process.lua: the training process.
* evaluate_process.lua: the evaluation process. 
* package 'model' contains the required models including attention model, TAGM, LSTM, GRU and plain-RNN. 
* package 'util' contains the required small utilities such as data loader.

__Data__   
The clean arabic data used in the paper is uploaded named 'data/arabic/window_2.t7'. You can generate the noised version by running 'util/arabic_preprocess.lua'. It will generate a noised version named 'arabic_append_noise.t7', which can be used in the main.lua for conducting the experiments for Speech Recognition Experiments in the paper.   
The datasets for the Sentiment analysis and Event recognition are too large to upload to github. They can be obtained from the original data source online.

Feel free to contact [Wenjie Pei](https://wenjiepei.github.io/) (wenjiecoder@gmail.com).

TAGM

TAGM Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

TAGM

TAGM Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

Download README