issues
search
Willu12
/
iml
MIT License
1
stars
0
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
a comparison to pre-trained models (transfer learning, fine-tuning)
#23
mytkom
opened
1 day ago
0
activation functions (ReLU, sigmoid) tested against diMerent weight initializations (Xavier, He, uniform)
#22
mytkom
opened
1 day ago
0
skip connections
#21
mytkom
opened
1 day ago
0
batch normalization (before vs after activations)
#20
mytkom
opened
1 day ago
0
different optimizers (SGD, Adam), learning rate schedule, weight decay
#19
mytkom
opened
1 day ago
0
changing the number and size of the layers
#18
mytkom
opened
1 day ago
0
removing silent passages from the training data
#17
mytkom
opened
1 day ago
0
Visual interpretation for CNN
#16
Willu12
opened
2 days ago
0
Added dropout and Monte Carlo Dropout
#15
Willu12
opened
3 days ago
0
Visual interpretation of cnn (interpretability feature from pdf).
#14
Willu12
opened
3 days ago
0
Added usage of pytorch.randomCrop instead of splitting clips.
#13
Willu12
opened
4 days ago
0
test dropout
#12
Willu12
opened
4 days ago
0
test monte carlo dropout
#11
Willu12
opened
4 days ago
0
saving best NN
#10
Willu12
closed
1 day ago
0
extend prepare_datasets logic to allow different data splitting methods
#9
bas0N
opened
2 weeks ago
2
(optional) change dataset and classification task
#8
mytkom
opened
3 weeks ago
0
2: normalization and standardization of spectrograms
#7
mytkom
closed
3 weeks ago
1
save the best NN instead of NN after max number of epochs
#6
mytkom
closed
1 day ago
0
Try to use PyTorch transform on long spectrograms to obtain shorter clips randomly
#5
mytkom
opened
3 weeks ago
0
Balance dataset
#4
mytkom
opened
3 weeks ago
0
Test dataset evaluation and statistics
#3
mytkom
opened
3 weeks ago
0
Normalization and standardization of spectrograms
#2
mytkom
closed
3 weeks ago
1
Project structure and simple CNN
#1
mytkom
closed
3 weeks ago
1