Set up a custom Dataset and DataLoader to load, transform, and batch data.
Split the complete dataset into Subsets for training, validation, and testing and witness the impact of Distribution Shift.
Explore why Architectural Bias in Convolutional Neural Networks acts as Automated Feature Engineering by learning Convolution Filters (or Kernels), allowing them to be more parameter-efficient compared to standard Multi-Layer Perceptrons.
Understand what Loss Surfaces are and how Gradient Descent and batch size influence model optimization.
MIT 6.S191: Introduction to Deep Learning This is a full course and absolutely optional. We recommed it as a resource after completing DeepLearning.AI's PyTorch for Deep Learning (feel free to use it at your dicretion, if you want another perspective on e.g., CNNs.)
3Blue1Brown's Neural Network Playlist (thanks for sharing!)
Cross Validation (StatQuest Video, 6 min) How to split the available data into subsets for training and testing and how Cross Validation is used to process different splits to evaluate your model's performance
Bias and Variance (Overfitting) (StatQuest, 7 min) The video discusses the Bias-Variance Tradeoff and explains why we want to design and train models that are at the sweet spot between
minimizing Bias (Underfitting) Error resulting from overly simple assumptions (the model misses the underlying pattern)
minimizing Variance (Overfitting) Error resulting from the model being too sensitive to the training data. High variance causes the model to "memorize" noise instead of learning the underlying distribution such that it fails to generalize on unseen data
Mini-Batch Gradient Descent helps us navigate this tradeoff by introducing just enough noise to prevent the high variance (overfitting) typically seen in Full-Batch Gradient Descent
Complete module 3 of the second course "PyTorch: Techniques and Ecosystem Tools"
Set up a repository for your project following the instructions given here
Conduct a literature review according to the instructions given here
Document your findings on the literature review according to the instructions of above
Complete the assignment in the following notebook:
You can find the helper functions required to run the notebook here. You can either load it the file into Colab or simply copy the code into a cell.
Last updated
Was this helpful?
Was this helpful?
