Deep Learning (Python) for Neuroscience EEG Practical course

Lecture 1: IntroductionHere you will find a short introduction to the course. We outline the objectives, structure, and practical outcomes. This sets the stage for a hands-on experience in machine learning with EEG signals.Lecture 2: Connect to Google ColabThis chapter provides a step-by-step guide on how to connect to and work in Google Colab. You’ll learn how to set up your environment, install required libraries, and ensure you are ready to run the code examples provided throughout the course.Lecture 3: Hardware for Brain-Computer InterfaceThis chapter covers the essential hardware used in EEG-based brain-computer interfaces. Lecture 4: Data EvaluationWe dive into evaluating the quality of your EEG data. This chapter explores techniques to inspect, clean, and annotate EEG recordings, ensuring that your data is reliable before moving forward with analysis or machine learning tasks.Lecture 5: Prepare the DatasetLearn how to transform raw EEG signals into structured datasets suitable for machine learning. This chapter includes labeling, segmenting, and feature extraction techniques—critical steps for successful model training and testing.Lecture 6: Introduction to DLIn this chapter, we introduce the fundamentals of deep learning and explain why Keras is a suitable library for working with EEG data. You’ll gain a basic understanding of deep learning concepts, how they apply to EEG signal processing, and where to find more information about Keras and its capabilities. This sets the foundation for implementing neural networks in upcoming lectures.Lecture 7. Convolutional Neural Networks (CNNs) for EEGThis chapter introduces convolutional neural networks (CNNs) and their application to EEG signal processing. You’ll learn the theory behind CNNs, how they are used for automatic feature extraction, and how to implement and fine-tune a CNN architecture for EEG data using Keras.Lecture 8. Recurrent Neural Networks (RNNs) and LSTMExplore how recurrent neural networks (RNNs), particularly Long Short-Term Memory (LSTM) networks, can model temporal dependencies in EEG signals. This chapter covers both the theoretical background and practical implementation, guiding you through the creation and optimization of LSTM architectures for EEG analysis.Lecture 9. Autoencoders and generative modelsDive into unsupervised deep learning with autoencoders and generative adversarial networks (GANs). Learn how these models can be used for feature learning, anomaly detection, and synthetic data generation in EEG applications. This chapter combines theory with hands-on examples using Keras.Lecture 10. ConclusionIn the final chapter, we summarize the key takeaways from the course and outline possible next steps for your learning journey.