training_config.py

import numpy as np
import tensorflow as tf


class train_Config(object):

    """Training Configurations"""
    max_grad_norm = 5                           # Gradient clipping threshold
    input_window_size = 50                      # Input window size during training
    output_window_size = 10                     # Output window size during training
    test_output_window = 100                    # Output window size during testing. test_output_window is overwritten by test set size when longterm is true
    hidden_size = 100                           # Number of hidden units for HMR
    batch_size = 16                             # Batch size for training
    learning_rate = 0.001                       # Learning rate
    max_epoch = 50                              # Maximum training epochs
    training_size = 500                         # Training iterations per epoch
    validation_size = 100                       # Validation iterations per epoch
    restore = False                             # Restore the trained weights or restart training from scratch
    longterm = False                            # Whether we are doing super longterm prediction
    early_stop = 10                             # Stop training if validation loss does not improve after these epochs
    keep_prob = 0.9                             # Keep probability for RNN cell weights
    context_window = 1                          # Context window size in HMR
    recurrent_steps = 3                         # Number of recurrent steps in HMR

    """Choice of model and loss function"""
    models = ['ERD', 'LSTM3lr', 'GRU', 'HMR']
    model = models[3]

    loss_funcs = ['l2', 'linearizedlie']
    loss = loss_funcs[0]

    def __init__(self, dataset, datatype, action):
        self.dataset = dataset
        self.datatype = datatype
        self.filename = action

        """Define kinematic chain configurations based on dataset class"""
        if self.dataset == 'Fish':
            self.filename = 'default'
            self.chain_config = [np.arange(0, 21)]
        elif self.dataset == 'Mouse':
            self.filename = 'default'
            self.chain_config = [np.arange(0, 5)]
        elif self.dataset == 'Human':
            self.chain_config = [np.array([0, 1, 2, 3, 4, 5]),
                                 np.array([0, 6, 7, 8, 9, 10]),
                                 np.array([0, 12, 13, 14, 15]),
                                 np.array([13, 17, 18, 19, 22, 19, 21]),
                                 np.array([13, 25, 26, 27, 30, 27, 29])]
            self.parents = np.array(
                [-1, 0, 1, 2, 3, 4, 0, 6, 7, 8, 9, 0, 11, 12, 13, 14, 12, 16, 17, 18, 19, 20, 19, 22, 12, 24, 25, 26,
                 27, 28, 27, 30])
            self.offsets = np.array([[0., 0., 0.],
                                     [-132.95, 0., 0.],
                                     [0., -442.89, 0.],
                                     [0., -454.21, 0.],
                                     [0., 0., 162.77],
                                     [0., 0., 75.],
                                     [132.95, 0., 0.],
                                     [0., -442.89, 0.],
                                     [0., -454.21, 0.],
                                     [0., 0., 162.77],
                                     [0., 0., 75.],
                                     [0., 0., 0.],
                                     [0., 233.38, 0.],
                                     [0., 257.08, 0.],
                                     [0., 121.13, 0.],
                                     [0., 115., 0.],
                                     [0., 257.08, 0.],
                                     [0., 151.03, 0.],
                                     [0., 278.88, 0.],
                                     [0., 251.73, 0.],
                                     [0., 0., 0.],
                                     [0., 0., 100.],
                                     [0., 137.5, 0.],
                                     [0., 0., 0.],
                                     [0., 257.08, 0.],
                                     [0., 151.03, 0.],
                                     [0., 278.88, 0.],
                                     [0., 251.73, 0.],
                                     [0., 0., 0.],
                                     [0., 0., 100.],
                                     [0., 137.5, 0.],
                                     [0., 0., 0.]])