Deep Learning Foundation is a three-day course that takes participants through foundational concepts, techniques, and applications in deep learning. Day one opens by tracing the relationship between AI, machine learning, and deep learning through real applications. It also covers the historical breakthroughs that shaped the field. Attendees then build neural network basics from the ground up, covering neurons, activation functions, feedforward architectures, and the backpropagation algorithm that drives learning. Model architecture design and data preprocessing follow. The day closes with training and evaluation, covering loss functions, evaluation metrics, and optimization techniques. Students also learn the regularization methods that keep models from overfitting.
Day two moves into Convolutional Neural Networks, starting with the mathematics of convolution. Attendees then apply CNN architectures and transfer learning to computer vision tasks such as image classification and object detection. Attendees then turn to Recurrent Neural Networks, including LSTM and GRU architectures. The day finishes with Transformer architectures, covering scaled dot product attention, attention masks, and how to put pretrained transformers to work.
Day three applies these techniques to natural language processing, covering text preprocessing, embeddings, and sequence modeling. It then moves into deployment and optimization, comparing local and cloud-based strategies and tuning models for inference speed and production monitoring. A closing block covers hyperparameter tuning, adversarial robustness, and the ethics of building production deep learning systems. A look at emerging trends closes out the week. This hands-on course spends at least half its time in lab sessions, and attendees leave prepared to work on real world deep learning projects.
Who Should Attend
Data Scientists, Data Engineers, AI/ML/MLOps Engineers, Programmers
What Attendees Will Learn
Upon completing Deep Learning Foundation, participants will be able to:
- Explain neural network fundamentals, including activation functions and backpropagation
- Design, build, and train neural network models with PyTorch
- Build and train Convolutional Neural Networks for computer vision tasks
- Build and train Recurrent Neural Networks, including LSTM and GRU architectures
- Implement Transformer architectures and apply them to NLP tasks
- Deploy, monitor, and optimize Deep Learning models for production
- Apply hyperparameter optimization and address adversarial robustness and ethics
Prerequisites
Participants should have programming skills, preferably in Python, an understanding of mathematics and statistics, and prior experience with machine learning concepts.