Exploring Edge Impulse and TinyML (Day 2 - Test 2)

What is the role of the SGD step in the training loop illustrated in the diagram?

• Make a prediction using weights and biases

• Measure the model's accuracy using MSE

• Optimize the model by updating weights and biases

• Calculate the number of epochs needed

What is Edge Impulse primarily used for?

• Designing websites for e-commerce.

• Managing cloud storage solutions.

• Developing and deploying machine learning models for edge devices.

• Creating video games for mobile devices.

Describe the process of developing a model in Edge Impulse.

• Deploying the model before training

• Collecting data without labeling it

• Skipping performance evaluation after training

• The process of developing a model in Edge Impulse involves creating a project, collecting and labeling data, configuring processing blocks, training the model, evaluating its performance, and deploying it.

What is TinyML and how does it relate to Edge Impulse?

• TinyML is the application of machine learning on small devices, and Edge Impulse is a platform that supports the development and deployment of TinyML models.

• TinyML is a type of hardware used for data storage.

• Edge Impulse is a cloud service for big data analytics.

• TinyML is exclusively for high-performance computing systems.

List the key steps in training a TinyML model using Edge Impulse.

• Create a model without data;

• Deploy model before training;

• Skip performance evaluation;

• 1. Create an account and project; 2. Collect and upload data; 3. Label data; 4. Select algorithm; 5. Train model; 6. Evaluate performance; 7. Optimize model; 8. Deploy to edge device.

What metrics are commonly used to evaluate model performance?

• Speed, Efficiency, Cost

• User Satisfaction, Feedback Score

• Data Size, Complexity, Training Time

• Accuracy, Precision, Recall, F1 Score, Mean Squared Error, Mean Absolute Error

Explain the significance of confusion matrix in model evaluation.

• The confusion matrix is significant as it provides a comprehensive view of a model's performance, enabling the calculation of key metrics and insights into prediction errors.

• It replaces the need for cross-validation.

• It is used to visualize the training data.

• It only shows the accuracy of the model.

What role does data preprocessing play in model development?

• Data preprocessing has no impact on model accuracy.

• Data preprocessing is primarily used for data visualization.

• Data preprocessing is only necessary for large datasets.

• Data preprocessing improves data quality and prepares it for effective model training.

What are the advantages of using Edge Impulse for ML model development?

• Complex setup process

• Limited support for edge devices

• No tools for data collection

• Advantages of using Edge Impulse include ease of use, support for diverse data types, integrated tools for data collection and model evaluation, and optimized deployment for edge devices.

According to the diagram, what connects all steps in the training process and is responsible for iterative improvement?

• Epochs (Back-Propagation)

• Loss Function

• Learning Rate

• Bias Terms