LSTM & Attention Mech…

The attention mechanism, initially used in natural language processing fields, has found its way into finance and other domains. It operates on a simple concept: some parts of the input sequence are more important then others. The attention mechanism improve models and understand's capabilities, by allowing the model to focus on specific parts of the input sequence while ignoring others.

Incorporating attention into LSTM networks give a context to the model for predictions, certain historical data points may be more relevant than others. The attention mechanism give the LSTM ability to weigh these points more heavily, leading to more accurate and nuanced predictions.

Prepare de Data ...


from keras.models import Sequential
from keras.layers import LSTM, Dense, Dropout, AdditiveAttention, Permute, Reshape, Multiply

model = Sequential()

model.add(LSTM(units=50, return_sequences=True, input_shape=(X_train.shape[1], 1)))
model.add(LSTM(units=50, return_sequences=True, name="lstm_layer"))
from keras.models import Sequential
from keras.layers import LSTM, Dense, Dropout, AdditiveAttention, Permute, Reshape, Multiply

model = Sequential()

# Adding LSTM layers with return_sequences=True
model.add(LSTM(units=50, return_sequences=True, input_shape=(X_train.shape[1], 1)))
model.add(LSTM(units=50, return_sequences=True, name="lstm_layer"))


attention = AdditiveAttention()

attention_result = attention([model.outputs[0], model.outputs[0]])
attention_result = Multiply()([model.outputs[0], attention_result])

model.add(tf.keras.layers.Flatten())
model.add(Dense(1))
attention = AdditiveAttention()

attention_result = attention([model.outputs[0], model.outputs[0]])
multiply_layer = Multiply()([model.outputs[0], attention_result])

model.add(tf.keras.layers.Flatten())
model.add(Dense(1))

Add the Dropout and BatchNormalization before Compiling ...

Train, Predict and Visualize

4nkh Development Team

Jun

DAPHNE service

Create the service file:


sudo vim /etc/systemd/system/daphne.service
sudo vim /etc/systemd/system/daphne.service


[Unit]
Description=daphne daemon
Requires=daphne.socket
After=network.target
[Service]
Type=simple
User=user_name
WorkingDirectory=/var/www/app_name
ExecStart=/home/user_name/python_env/bin/daphne -u /tmp/daphne.sock app_name.asgi:application
[Install]
WantedBy=multi-user.target
[Unit]
Description=daphne daemon
Requires=daphne.socket
After=network.target
[Service]
Type=simple
User=user_name
WorkingDirectory=/var/www/app_name
ExecStart=/home/user_name/python_env/bin/daphne -u /tmp/daphne.sock app_name.asgi:application
[Install]
WantedBy=multi-user.target

Command lines:


sudo systemctl start daphne.service
sudo systemctl stop daphne.service
sudo systemctl restart daphne.service
sudo systemctl status daphne.service
sudo systemctl start daphne.service
sudo systemctl stop daphne.service
sudo systemctl restart daphne.service
sudo systemctl status daphne.service

4nkh Development Team

May

Change ssh port

Change the specified port:

vim /etc/ssh/sshd_config
vim /etc/ssh/sshd_config

Restart the deamon:

systemctl restart ssh
service ssh restart
systemctl restart ssh
service ssh restart

Change the port and verify:

ss -tlpn| grep ssh
netstat -tlpn| grep ssh
ss -tlpn| grep ssh
netstat -tlpn| grep ssh

Ask access to firewall:

ufw allow 30/tcp
ufw allow 30/tcp

If you're on a VPS with a firewall, give access to the port

4nkh Development Team

May

Unix startup script

Build your unix script:

vim /home/user_name/unix_script/python_script.sh
vim /home/user_name/unix_script/python_script.sh

python_script.sh:

#!/bin/bash

echo "get python version"
python3 --version
#!/bin/bash

echo "get python version"
python3 --version

Build the service to run the script:

vim /lib/systemd/system/startup.service startup.service
vim /lib/systemd/system/startup.service

startup.service:

[Unit]
Description=Startup Script
[Service]
ExecStart=/bin/bash "/home/user_name/unix_script/python_script.sh"
[Install]
WantedBy=multi-user.target
[Unit]
Description=Startup Script
[Service]
ExecStart=/bin/bash "/home/user_name/unix_script/python_script.sh"
[Install]
WantedBy=multi-user.target

Enable the service and build the symlink:

systemctl enable startup.service --now
systemctl enable startup.service --now

4nkh Development Team

May