Repository of code used to generate the search results for the article "Semantic malware classification using convolutional neural networks".
The repository does not contain malware samples.
The instructions and command suggestions are examples of use for generating results compatible with those presented in the article.
Comments:
- Some code snippets may not have had the correct indication of the author, please let me know if I used codes, figures, graphics and did not cite the author correctly;
- Some comments, variable names and methods may be written in Portuguese or Spanish;
- GPU usage is required for full testing, if one is not available some warnings may be displayed;
Install Python (https://www.python.org/), pip (https://pypi.org/project/pip/) e Conda (https://www.anaconda.com/products/distribution).
wget -c https://repo.anaconda.com/archive/Anaconda3-2021.11-Linux-x86_64.sh
bash Anaconda3-2021.11-Linux-x86_64.sh
Create a python virtual environment with Anaconda for the tests:
conda create --name semantic-pe-cnn python=3.9 anaconda -y
conda activate semantic-pe-cnn
conda install -c anaconda py-lief==0.10.1 -y
conda install -c conda-forge imbalanced-learn -y
pip install tensorflow==2.9.0 keras==2.9.0 livelossplot
Malware samples were provided by Ricardo Sant'Ana (https://github.com/ricksant2003/MalwareDatasetVirusShareSant).
To generate the sample data you need to request the binary files with the malware samples from the author of VirusShareSant, and unpack the archive contents into the directory "VirusShareSant/pe-malwares".
Then run the following commands to separate the files into semantic parts according to byte amounts used in the article's proofs:
python process_pe_part.py --num_samples=0 --pe_part=header --cleanup
python process_pe_part.py --pe_part=code --pe_size=100000 --num_samples=0 --cleanup
python process_pe_part.py --pe_part=data --pe_size=100000 --num_samples=0 --cleanup
python process_pe_part.py --pe_part=file --pe_size=250000 --num_samples=0 --cleanup
The results will be stored in the folder VirusShareSant/pe-parts/{header,code,data,file}.
To train the CNN models, test, evaluate and collect the results, the following commands must be executed:
python train.py --train_memory --verbose=1 --rounds=6 --pe_part=header --pe_size=4096 --num_samples=0
python train.py --train_memory --verbose=1 --rounds=6 --pe_part=file --pe_size=5000 --pe_increment=5000 --num_samples=0
python train.py --train_memory --verbose=1 --rounds=6 --pe_part=file --pe_size=5000 --pe_increment=5000 --num_samples=0
python train.py --train_memory --verbose=1 --rounds=6 --pe_part=file --pe_size=5000 --pe_increment=5000 --num_samples=0
To view the results run the commands below:
python results.py --pe_part=header
python results.py --pe_part=code
python results.py --pe_part=data
python results.py --pe_part=file
Just for the purpose of a more complete example, the following is a list of executed commands and result previews of training examples with 100 samples of the semantic part headers. The tests and data provided here do not represent any results used in the article.
Extract semantic header part from PE files:
(semantic-pe-cnn) $ python process_pe_part.py --num_samples=0 --pe_part=header --cleanup
--------------------------------------------------------------------------------
Semantic PE Malware Classifier (SPEMC) - Process PE Files
--------------------------------------------------------------------------------
Time Start: 2022-12-06 19:47:57.014530
Removing all the existing processed files
Loading original malware metadata from VirusShareSant...
X_exp: (28617,)
y_exp: (28617,)
Not balanced...
Parallel processing files
Time End: 2022-12-06 20:01:15.430244 - 0:13:18.398213
Training:
(semantic-pe-cnn) $ python train.py --train_memory --verbose=1 --rounds=6 --pe_part=header --pe_size=4096 --num_samples=100
--------------------------------------------------------------------------------
Semantic PE Malware Classifier (SPEMC) - Training
--------------------------------------------------------------------------------
Start the training: 2022-12-06 20:29:48.264245
Start Round 1: 2022-12-06 20:29:48.264315
Load data to memory...
Time Start: 2022-12-06 20:29:52.302580
Loading original malware metadata from VirusShareSant...
X_exp: (28617,)
y_exp: (28617,)
Over sampling dataset...
X_over: (61146, 1)
y_over: (61146, 1)
Splitting Data into Training, Validation, and Testing...
mum_samples: 100
train: (80, 2)
test: (10, 2)
val: (10, 2)
Load train data
header - Size: 320.00KB , Shape: (80, 4096)
Load val data
header - Size: 40.00KB , Shape: (10, 4096)
Load test data
header - Size: 40.00KB , Shape: (10, 4096)
Samples data:
X_train: (80, 4096)
y_train: (80,)
X_val: (10, 4096)
y_val: (10,)
X_test: (10, 4096)
y_test: (10,)
Time End: 2022-12-06 20:29:56.936385 - 0:00:04.633787
\Model fit...
Epoch 1/1000
3/3 [==============================] - 3s 782ms/step - loss: 2.1984 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1973 - val_sparse_categorical_accuracy: 0.2000
Epoch 2/1000
3/3 [==============================] - 2s 617ms/step - loss: 2.1945 - sparse_categorical_accuracy: 0.1750 - val_loss: 2.1934 - val_sparse_categorical_accuracy: 0.2000
Epoch 3/1000
3/3 [==============================] - 2s 565ms/step - loss: 2.1918 - sparse_categorical_accuracy: 0.1750 - val_loss: 2.1914 - val_sparse_categorical_accuracy: 0.2000
Epoch 4/1000
3/3 [==============================] - 2s 573ms/step - loss: 2.1879 - sparse_categorical_accuracy: 0.1750 - val_loss: 2.1914 - val_sparse_categorical_accuracy: 0.2000
Epoch 5/1000
3/3 [==============================] - 2s 579ms/step - loss: 2.1844 - sparse_categorical_accuracy: 0.1750 - val_loss: 2.1895 - val_sparse_categorical_accuracy: 0.2000
Epoch 6/1000
3/3 [==============================] - 2s 599ms/step - loss: 2.1809 - sparse_categorical_accuracy: 0.1750 - val_loss: 2.1875 - val_sparse_categorical_accuracy: 0.2000
Epoch 7/1000
3/3 [==============================] - 2s 601ms/step - loss: 2.1770 - sparse_categorical_accuracy: 0.1750 - val_loss: 2.1875 - val_sparse_categorical_accuracy: 0.2000
Epoch 8/1000
3/3 [==============================] - 2s 574ms/step - loss: 2.1730 - sparse_categorical_accuracy: 0.1750 - val_loss: 2.1836 - val_sparse_categorical_accuracy: 0.2000
Epoch 9/1000
3/3 [==============================] - 2s 605ms/step - loss: 2.1703 - sparse_categorical_accuracy: 0.1750 - val_loss: 2.1836 - val_sparse_categorical_accuracy: 0.2000
Epoch 10/1000
3/3 [==============================] - 2s 566ms/step - loss: 2.1660 - sparse_categorical_accuracy: 0.1750 - val_loss: 2.1836 - val_sparse_categorical_accuracy: 0.2000
Epoch 11/1000
3/3 [==============================] - 2s 583ms/step - loss: 2.1629 - sparse_categorical_accuracy: 0.1750 - val_loss: 2.1836 - val_sparse_categorical_accuracy: 0.2000
Epoch 11: early stopping
Model evaluate...
1/1 [==============================] - 0s 26ms/step - loss: 2.2539 - sparse_categorical_accuracy: 0.1000
Model predict...
1/1 [==============================] - 0s 123ms/step
Store trains and tests results...
End round 1: 2022-12-06 20:30:19.841992 - 0:00:31.577668
--------------------------------------------------------------------------------
Start Round 2: 2022-12-06 20:30:19.842067
Load data to memory...
Time Start: 2022-12-06 20:30:19.916479
Loading original malware metadata from VirusShareSant...
X_exp: (28617,)
y_exp: (28617,)
Over sampling dataset...
X_over: (61146, 1)
y_over: (61146, 1)
Splitting Data into Training, Validation, and Testing...
mum_samples: 100
train: (80, 2)
test: (10, 2)
val: (10, 2)
Load train data
header - Size: 320.00KB , Shape: (80, 4096)
Load val data
header - Size: 40.00KB , Shape: (10, 4096)
Load test data
header - Size: 40.00KB , Shape: (10, 4096)
Samples data:
X_train: (80, 4096)
y_train: (80,)
X_val: (10, 4096)
y_val: (10,)
X_test: (10, 4096)
y_test: (10,)
Time End: 2022-12-06 20:30:20.536588 - 0:00:00.620067
\Model fit...
Epoch 1/1000
3/3 [==============================] - 3s 692ms/step - loss: 2.1969 - sparse_categorical_accuracy: 0.1000 - val_loss: 2.1973 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 2/1000
3/3 [==============================] - 2s 605ms/step - loss: 2.1953 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.1934 - val_sparse_categorical_accuracy: 0.2000
Epoch 3/1000
3/3 [==============================] - 2s 579ms/step - loss: 2.1934 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1914 - val_sparse_categorical_accuracy: 0.2000
Epoch 4/1000
3/3 [==============================] - 2s 577ms/step - loss: 2.1910 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.1895 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 5/1000
3/3 [==============================] - 2s 554ms/step - loss: 2.1891 - sparse_categorical_accuracy: 0.2250 - val_loss: 2.1875 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 6/1000
3/3 [==============================] - 2s 575ms/step - loss: 2.1859 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1875 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 7/1000
3/3 [==============================] - 2s 578ms/step - loss: 2.1855 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1875 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 8/1000
3/3 [==============================] - 2s 576ms/step - loss: 2.1824 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1875 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 9/1000
3/3 [==============================] - 2s 555ms/step - loss: 2.1809 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1855 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 10/1000
3/3 [==============================] - 2s 557ms/step - loss: 2.1785 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1875 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 11/1000
3/3 [==============================] - 2s 567ms/step - loss: 2.1770 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1855 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 12/1000
3/3 [==============================] - 2s 637ms/step - loss: 2.1750 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1836 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 12: early stopping
Model evaluate...
1/1 [==============================] - 0s 35ms/step - loss: 2.1836 - sparse_categorical_accuracy: 0.1000
Model predict...
1/1 [==============================] - 0s 138ms/step
Store trains and tests results...
End round 2: 2022-12-06 20:30:44.803190 - 0:00:24.961109
--------------------------------------------------------------------------------
Start Round 3: 2022-12-06 20:30:44.803269
Load data to memory...
Time Start: 2022-12-06 20:30:44.890833
Loading original malware metadata from VirusShareSant...
X_exp: (28617,)
y_exp: (28617,)
Over sampling dataset...
X_over: (61146, 1)
y_over: (61146, 1)
Splitting Data into Training, Validation, and Testing...
mum_samples: 100
train: (80, 2)
test: (10, 2)
val: (10, 2)
Load train data
header - Size: 320.00KB , Shape: (80, 4096)
Load val data
header - Size: 40.00KB , Shape: (10, 4096)
Load test data
header - Size: 40.00KB , Shape: (10, 4096)
Samples data:
X_train: (80, 4096)
y_train: (80,)
X_val: (10, 4096)
y_val: (10,)
X_test: (10, 4096)
y_test: (10,)
Time End: 2022-12-06 20:30:45.569804 - 0:00:00.678955
\Model fit...
Epoch 1/1000
3/3 [==============================] - 3s 681ms/step - loss: 2.1977 - sparse_categorical_accuracy: 0.0625 - val_loss: 2.2012 - val_sparse_categorical_accuracy: 0.1000
Epoch 2/1000
3/3 [==============================] - 2s 579ms/step - loss: 2.1965 - sparse_categorical_accuracy: 0.0750 - val_loss: 2.1992 - val_sparse_categorical_accuracy: 0.1000
Epoch 3/1000
3/3 [==============================] - 2s 593ms/step - loss: 2.1953 - sparse_categorical_accuracy: 0.1375 - val_loss: 2.1992 - val_sparse_categorical_accuracy: 0.1000
Epoch 4/1000
3/3 [==============================] - 3s 814ms/step - loss: 2.1941 - sparse_categorical_accuracy: 0.1375 - val_loss: 2.1992 - val_sparse_categorical_accuracy: 0.1000
Epoch 5/1000
3/3 [==============================] - 2s 661ms/step - loss: 2.1918 - sparse_categorical_accuracy: 0.1375 - val_loss: 2.1992 - val_sparse_categorical_accuracy: 0.1000
Epoch 6/1000
3/3 [==============================] - 2s 565ms/step - loss: 2.1914 - sparse_categorical_accuracy: 0.1375 - val_loss: 2.1992 - val_sparse_categorical_accuracy: 0.1000
Epoch 7/1000
3/3 [==============================] - 2s 621ms/step - loss: 2.1895 - sparse_categorical_accuracy: 0.1750 - val_loss: 2.1992 - val_sparse_categorical_accuracy: 0.1000
Epoch 8/1000
3/3 [==============================] - 3s 899ms/step - loss: 2.1883 - sparse_categorical_accuracy: 0.2250 - val_loss: 2.1992 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 9/1000
3/3 [==============================] - 2s 689ms/step - loss: 2.1855 - sparse_categorical_accuracy: 0.2500 - val_loss: 2.1992 - val_sparse_categorical_accuracy: 0.1000
Epoch 10/1000
3/3 [==============================] - 2s 561ms/step - loss: 2.1840 - sparse_categorical_accuracy: 0.2750 - val_loss: 2.1973 - val_sparse_categorical_accuracy: 0.1000
Epoch 11/1000
3/3 [==============================] - 2s 703ms/step - loss: 2.1840 - sparse_categorical_accuracy: 0.2500 - val_loss: 2.1992 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 11: early stopping
Model evaluate...
1/1 [==============================] - 0s 34ms/step - loss: 2.1816 - sparse_categorical_accuracy: 0.1000
Model predict...
1/1 [==============================] - 0s 122ms/step
Store trains and tests results...
End round 3: 2022-12-06 20:31:11.111717 - 0:00:26.308438
--------------------------------------------------------------------------------
Start Round 4: 2022-12-06 20:31:11.111784
Load data to memory...
Time Start: 2022-12-06 20:31:11.206678
Loading original malware metadata from VirusShareSant...
X_exp: (28617,)
y_exp: (28617,)
Over sampling dataset...
X_over: (61146, 1)
y_over: (61146, 1)
Splitting Data into Training, Validation, and Testing...
mum_samples: 100
train: (80, 2)
test: (10, 2)
val: (10, 2)
Load train data
header - Size: 320.00KB , Shape: (80, 4096)
Load val data
header - Size: 40.00KB , Shape: (10, 4096)
Load test data
header - Size: 40.00KB , Shape: (10, 4096)
Samples data:
X_train: (80, 4096)
y_train: (80,)
X_val: (10, 4096)
y_val: (10,)
X_test: (10, 4096)
y_test: (10,)
Time End: 2022-12-06 20:31:11.945289 - 0:00:00.738594
\Model fit...
Epoch 1/1000
3/3 [==============================] - 3s 652ms/step - loss: 2.1992 - sparse_categorical_accuracy: 0.0750 - val_loss: 2.1992 - val_sparse_categorical_accuracy: 0.2000
Epoch 2/1000
3/3 [==============================] - 2s 569ms/step - loss: 2.1953 - sparse_categorical_accuracy: 0.1000 - val_loss: 2.2070 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 3/1000
3/3 [==============================] - 2s 774ms/step - loss: 2.1910 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.2148 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 4/1000
3/3 [==============================] - 3s 763ms/step - loss: 2.1867 - sparse_categorical_accuracy: 0.1750 - val_loss: 2.2227 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 5/1000
3/3 [==============================] - 2s 737ms/step - loss: 2.1816 - sparse_categorical_accuracy: 0.1875 - val_loss: 2.2305 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 6/1000
3/3 [==============================] - 2s 573ms/step - loss: 2.1781 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.2402 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 7/1000
3/3 [==============================] - 2s 746ms/step - loss: 2.1734 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.2480 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 8/1000
3/3 [==============================] - 3s 835ms/step - loss: 2.1691 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.2578 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 9/1000
3/3 [==============================] - 3s 774ms/step - loss: 2.1645 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.2676 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 10/1000
3/3 [==============================] - 2s 723ms/step - loss: 2.1609 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.2773 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 11/1000
3/3 [==============================] - 2s 582ms/step - loss: 2.1574 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.2852 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 11: early stopping
Model evaluate...
1/1 [==============================] - 0s 27ms/step - loss: 2.2227 - sparse_categorical_accuracy: 0.1000
Model predict...
1/1 [==============================] - 0s 96ms/step
Store trains and tests results...
End round 4: 2022-12-06 20:31:38.601964 - 0:00:27.490171
--------------------------------------------------------------------------------
Start Round 5: 2022-12-06 20:31:38.602024
Load data to memory...
Time Start: 2022-12-06 20:31:38.679811
Loading original malware metadata from VirusShareSant...
X_exp: (28617,)
y_exp: (28617,)
Over sampling dataset...
X_over: (61146, 1)
y_over: (61146, 1)
Splitting Data into Training, Validation, and Testing...
mum_samples: 100
train: (80, 2)
test: (10, 2)
val: (10, 2)
Load train data
header - Size: 320.00KB , Shape: (80, 4096)
Load val data
header - Size: 40.00KB , Shape: (10, 4096)
Load test data
header - Size: 40.00KB , Shape: (10, 4096)
Samples data:
X_train: (80, 4096)
y_train: (80,)
X_val: (10, 4096)
y_val: (10,)
X_test: (10, 4096)
y_test: (10,)
Time End: 2022-12-06 20:31:39.358037 - 0:00:00.678205
\Model fit...
Epoch 1/1000
3/3 [==============================] - 4s 954ms/step - loss: 2.1980 - sparse_categorical_accuracy: 0.0750 - val_loss: 2.1973 - val_sparse_categorical_accuracy: 0.0000e+00
Epoch 2/1000
3/3 [==============================] - 3s 787ms/step - loss: 2.1957 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1973 - val_sparse_categorical_accuracy: 0.1000
Epoch 3/1000
3/3 [==============================] - 3s 796ms/step - loss: 2.1945 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.1973 - val_sparse_categorical_accuracy: 0.1000
Epoch 4/1000
3/3 [==============================] - 3s 771ms/step - loss: 2.1937 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.1973 - val_sparse_categorical_accuracy: 0.1000
Epoch 5/1000
3/3 [==============================] - 2s 580ms/step - loss: 2.1918 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.1973 - val_sparse_categorical_accuracy: 0.1000
Epoch 6/1000
3/3 [==============================] - 2s 767ms/step - loss: 2.1895 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.1973 - val_sparse_categorical_accuracy: 0.1000
Epoch 7/1000
3/3 [==============================] - 3s 781ms/step - loss: 2.1879 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.1973 - val_sparse_categorical_accuracy: 0.1000
Epoch 8/1000
3/3 [==============================] - 2s 766ms/step - loss: 2.1863 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.1992 - val_sparse_categorical_accuracy: 0.1000
Epoch 9/1000
3/3 [==============================] - 3s 775ms/step - loss: 2.1836 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.1992 - val_sparse_categorical_accuracy: 0.1000
Epoch 10/1000
3/3 [==============================] - 2s 728ms/step - loss: 2.1816 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.1992 - val_sparse_categorical_accuracy: 0.1000
Epoch 11/1000
3/3 [==============================] - 2s 579ms/step - loss: 2.1820 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.2012 - val_sparse_categorical_accuracy: 0.1000
Epoch 12/1000
3/3 [==============================] - 2s 767ms/step - loss: 2.1797 - sparse_categorical_accuracy: 0.1625 - val_loss: 2.2031 - val_sparse_categorical_accuracy: 0.1000
Epoch 12: early stopping
Model evaluate...
1/1 [==============================] - 0s 33ms/step - loss: 2.2227 - sparse_categorical_accuracy: 0.1000
Model predict...
1/1 [==============================] - 0s 129ms/step
Store trains and tests results...
End round 5: 2022-12-06 20:32:09.572688 - 0:00:30.970650
--------------------------------------------------------------------------------
Start Round 6: 2022-12-06 20:32:09.572765
Load data to memory...
Time Start: 2022-12-06 20:32:09.675283
Loading original malware metadata from VirusShareSant...
X_exp: (28617,)
y_exp: (28617,)
Over sampling dataset...
X_over: (61146, 1)
y_over: (61146, 1)
Splitting Data into Training, Validation, and Testing...
mum_samples: 100
train: (80, 2)
test: (10, 2)
val: (10, 2)
Load train data
header - Size: 320.00KB , Shape: (80, 4096)
Load val data
header - Size: 40.00KB , Shape: (10, 4096)
Load test data
header - Size: 40.00KB , Shape: (10, 4096)
Samples data:
X_train: (80, 4096)
y_train: (80,)
X_val: (10, 4096)
y_val: (10,)
X_test: (10, 4096)
y_test: (10,)
Time End: 2022-12-06 20:32:10.424919 - 0:00:00.749621
\Model fit...
Epoch 1/1000
3/3 [==============================] - 4s 913ms/step - loss: 2.1969 - sparse_categorical_accuracy: 0.1750 - val_loss: 2.1934 - val_sparse_categorical_accuracy: 0.4000
Epoch 2/1000
3/3 [==============================] - 2s 579ms/step - loss: 2.1953 - sparse_categorical_accuracy: 0.1875 - val_loss: 2.1934 - val_sparse_categorical_accuracy: 0.4000
Epoch 3/1000
3/3 [==============================] - 2s 764ms/step - loss: 2.1918 - sparse_categorical_accuracy: 0.1750 - val_loss: 2.1914 - val_sparse_categorical_accuracy: 0.4000
Epoch 4/1000
3/3 [==============================] - 2s 755ms/step - loss: 2.1875 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1895 - val_sparse_categorical_accuracy: 0.4000
Epoch 5/1000
3/3 [==============================] - 2s 761ms/step - loss: 2.1840 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1836 - val_sparse_categorical_accuracy: 0.4000
Epoch 6/1000
3/3 [==============================] - 2s 757ms/step - loss: 2.1797 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1816 - val_sparse_categorical_accuracy: 0.4000
Epoch 7/1000
3/3 [==============================] - 2s 761ms/step - loss: 2.1750 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1777 - val_sparse_categorical_accuracy: 0.4000
Epoch 8/1000
3/3 [==============================] - 2s 718ms/step - loss: 2.1719 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1758 - val_sparse_categorical_accuracy: 0.4000
Epoch 9/1000
3/3 [==============================] - 2s 595ms/step - loss: 2.1684 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1738 - val_sparse_categorical_accuracy: 0.4000
Epoch 10/1000
3/3 [==============================] - 2s 762ms/step - loss: 2.1652 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1719 - val_sparse_categorical_accuracy: 0.4000
Epoch 11/1000
3/3 [==============================] - 2s 757ms/step - loss: 2.1633 - sparse_categorical_accuracy: 0.1500 - val_loss: 2.1699 - val_sparse_categorical_accuracy: 0.4000
Epoch 11: early stopping
Model evaluate...
1/1 [==============================] - 0s 33ms/step - loss: 2.2031 - sparse_categorical_accuracy: 0.1000
Model predict...
1/1 [==============================] - 0s 119ms/step
Store trains and tests results...
End round 6: 2022-12-06 20:32:38.095274 - 0:00:28.522499
--------------------------------------------------------------------------------
End training: 2022-12-06 20:32:38.095340 - 0:02:49.831092
Results:
(semantic-pe-cnn) $ python results.py --pe_part=header
--------------------------------------------------------------------------------
Semantic PE Malware Classifier (SPEMC) - Results
--------------------------------------------------------------------------------
Results for part: header
0 train_data_men/header/2022-12-06-20-29-48
result_path in use: train_data_men/header/2022-12-06-20-29-48
--------------------------------------------------------------------------------
round imput_len params epoca loss acc val_loss val_acc eval_loss eval_acc pred % duration time
0 1 4096 215945 10 2.162891 17.500000 2.183594 20.000000 2.253906 10.0 10.0 00:00:31 31.544308
1 2 4096 215945 11 2.175000 15.000001 2.183594 0.000000 2.183594 10.0 10.0 00:00:24 24.929847
2 3 4096 215945 10 2.183984 25.000000 2.199219 0.000000 2.181641 10.0 10.0 00:00:26 26.277617
3 4 4096 215945 10 2.157422 16.249999 2.285156 0.000000 2.222656 10.0 10.0 00:00:27 27.461007
4 5 4096 215945 11 2.179688 16.249999 2.203125 10.000000 2.222656 10.0 10.0 00:00:30 30.937571
5 6 4096 215945 10 2.163281 15.000001 2.169922 40.000001 2.203125 10.0 10.0 00:00:28 28.491741
media 0 4096 215945 10 2.170378 17.500000 2.204102 11.666667 2.211263 10.0 10.0 00:00:28 28.273682
The results are storaged in the folder train_data_men/header/2022-12-06-20-29-48.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent