Home / IT & Computer Science / Biology & Biotechnology / Deep Learning for Bioscientists / Week 3 summary
Week 3 summary
A summary of week three of the course deep learning for bioscientists, on the training loop, hyperparametersm datasets and dataloaders.
Well done – we are now more than halfway through the course!
By now, we hope you have a basic working understanding of what convolutional neural networks are, how they are trained, and the types of data that are used in them.
In this week we have covered:
- the training loop, including:
- network output and loss functions
- how to make a training loop in PyTorch
- adusting hyperparameters such as learning rate and batch size
- datasets and dataloaders – software objects that help with managing your data and streamlining performance.
Next week, we will put all we have learned so far together and build a working image classifier using a publicly available image dataset depicting different flower species. After this, we will then see how we can adapt the network to perform a regression task, rather than a classification task.
