class Hyperparameters:
INPUT_LAYER = 43
HIDDEN_LAYER = 75 #Modify??
OUTPUT_LAYER = 3
NUM_EPOCHS = 10000
#NUM_EPOCHS = 1
BATCH_NUMBER = 240
LEARNING_RATE = 0.1
VALIDATION_NUMBER = 30
TEST_NUMBER = 30
HYP = Hyperparameters()
set_maker = SM()
pbfilename = "GraphV5/GRAPHS/GraphV5_frozen.pb"
file_name = "sen_data/inhale/61.wav"
prediction_dictionary = {0: "inhale", 1: "exhale", 2: "unknown"}
with tf.gfile.GFile(pbfilename, "rb") as f:
graph_def = tf.GraphDef()
graph_def.ParseFromString(f.read())
with tf.Graph().as_default() as graph:
tf.import_graph_def(graph_def,
input_map=None,
return_elements=None,
name="")
input = graph.get_tensor_by_name("placeholders/input_placeholder:0")
output = graph.get_tensor_by_name("prediction_and_loss/output:0")
NUM_EPOCHS = 10001
#NUM_EPOCHS = 1
BATCH_NUMBER = 340
LEARNING_RATE = 0.1
VALIDATION_NUMBER = 30
TEST_NUMBER = 30
class Information:
INPUT_DIMENSIONS = 43
INPUT_TIME_DIV = 0.125
INPUT_SECTORS = 8
SAMPLE_RATE = 4096
set_maker = SM()
HYP = Hyperparameters()
prediction_dictionary = {0: "inhale", 1: "exhale", 2: "unknown"}
W_In = tf.Variable(
tf.random_normal(shape=[HYP.INPUT_LAYER, HYP.HIDDEN_LAYER],
stddev=0.1,
mean=0),
name="W_In") #note: this used to have a mean of zero, so check that
W_Hidd = tf.Variable(tf.random_normal(
shape=[HYP.HIDDEN_LAYER, HYP.HIDDEN_LAYER], stddev=0.1, mean=0),
name="W_Hidd")
W_Out = tf.Variable(tf.random_normal(
shape=[HYP.HIDDEN_LAYER, HYP.OUTPUT_LAYER], stddev=0.1, mean=0),
name="W_Out")