Latest News

The AML Dashboard is a web-based tool that allows users to interact with the AML framework. Designed for developers looking to implement and test their own Machine Learning (AML) models, it provides an intuitive interface and advanced tools for model evaluation, fine-tuning, and validation across various scenarios. With this dashboard, users can monitor model performance, make real-time adjustments, and optimize deployment in production environments.

This tutorial showcases the training process of an AML model using the AML Dashboard.

Prerequisites

Ensure you have installed the AML Dashboard using one of the following methods:

• Linux Installation
• Docker Image Installation

For more information, check the AML Dashboard Interfaces and AML Dashboard Usage sections.

Running the demo

To run the necessary components for training a model using the AML Dashboard, follow these steps:

Start the backend server

1. Navigate to the backend directory.

2. Load the AML-IP environment.

3. Start the server:

Start the Computing Node

1. Load the AML-IP environment.

2. Navigate to the backend directory.

3. Start one or more computing nodes:

Each computing node will wait for job assignments and will collectively distribute the workload when multiple nodes are running.

Start the AML Dashboard

1. Navigate to the frontend/aml_dashboard directory.

2. Start the AML Dashboard:

3. Access the dashboard at http://localhost:5173/.

Training the Model

To train a model using the AML Dashboard, follow these steps:

1. Navigate to the Training tab on the AML Dashboard.
2. Specify the training parameters:
   • Number of parallel trainings (executions).
   • Number of iterations per execution.
   • Percentage of the dataset to distribute in each execution.
   • The target class when the classification is binary( this is the case when standard or custom datasets are used).
3. Configure the neural network parameters:
   • Number of layers.
   • Number of epochs.
   • Batch size.

This step is optional and intended for users who want to train a neural network as a benchmark to assess AML’s performance. By comparing the results of a neural network with those obtained using AML, users can gain insights into the effectiveness, accuracy, and efficiency of AML in handling their specific tasks. 

Note

The target class previously set will also be used for the neural network training.

4. Optionally, upload an atomization file as a pre-trained model.
5. Click on the Train button in the AML Training Launcher to initiate the training process.
6. The training progress will be displayed, and the model status will update to Finished :) once the training is completed.

Next Steps

In the following tutorial, we will guide you through the process of using a Collaborative Learning Scenario to fetch a trained model using AML-Dashboard.

By Denisa Alexandru from eProsima

MORE INFORMATION ABOUT ALMA:

For any questions please contact This email address is being protected from spambots. You need JavaScript enabled to view it..