Coordinated-Multi-Agent-Imitation-Learning

Coordinated-Multi-Agent-Imitation-Learning copied to clipboard

Would be good to see you after a while if you're still in Toronto. Here's my number 314 132 3417 (+1 to each digit)

ironyman

Where Can I find the soccer tracking dataset?

1

these 2 paper all use soccer tracking dataset, so where Can I find the soccer tracking dataset?

fenngMing

Hi, thank you for sharing your work. Could you explain how can I start training and run a trained model? Thank you,

moonsh

903124

Try MAP (Maximum a posteriori probability) instead of Viterbi

1

It may take too much time, but seems like MAP might produce a better result

samshipengs

Overfitting

2

> 2018-05-19 20:15:18,769 | INFO : Training with hyper parameters: {'use_model': 'dynamic_rnn_layer_norm', 'batch_size': 64, 'sequence_length': 50, 'overlap': 25, 'state_size': [128, 128], 'use_peepholes': None, 'input_dim': 179, 'dropout_rate': 0.6, 'learning_rate': 0.0001, 'n_epoch':...

samshipengs

https://arxiv.org/pdf/1606.00968.pdf http://www.yisongyue.com/publications/cvpr2016_online_smooth_long.pdf

samshipengs

Play by play data is not accurate

1

For example, In game `0021500196`, event `2`, `'time_left': [705, 704, 685, 684]}`, `'event_str': ['miss', 'rebound', 'miss', 'rebound'],` and match these with the shot clock left and the court visualization, the...

samshipengs