jdc08161063 / backchannel-prediction Goto Github PK

Code for my bachelor thesis and the corresponding paper

The final configurations are in configs/finunified. All of the ones in vary-* are generated with configs/meta_config_generator.ts

See data/README.md for more details.

cd janus
mkdir build && cd build
cmake ..
make -j$(nproc)
sudo python setup.py develop

You can reproduce the results of the paper using the script scripts/reproduce.sh.

Note that this may take a long time (~3h to extract the data (only once), ~2h to train one LSTM, 1h to evaluate it).

Generates configurations from a set of combinations

Run this from the project root:

ts-node --fast configs/meta_config_generator.ts

The best network configuration according to the objective evaluation is

configs/finunified/vary-features/lstm-best-features-power,pitch,ffv,word2vec_dim30.json

Server is in /web_vis/py/

Run this from the project root:

python -m web_vis.py.server extract/config.json

Client is in /web_vis/ts/

Run this from the folder /web_vis/ts/

yarn run dev

This will start a webserver serving the client at http://localhost:3000, which will connect to the server via websockets at localhost:8765.

The main script for extraction is extract/readDB.py. Run it via

export JOBS=4 # run in parallel
python -m extract.readDB configs/...

Note that the extraction will also be run automatically when before training when necessary, with all the results being cached. The data/cache folder will grow to about 10-20 GByte.

Build it and run the server

cd evaluate/plot
yarn
yarn run dev

Then go to http://localhost:8080/evaluate/plot/dist/

You can see more information in Section 6: Implementation of my bachelor's thesis, see here: https://github.com/phiresky/bachelor-thesis/blob/master/build/thesis.pdf