GPU-enabled workspaces (100 hours time limit).Project: Deploying a Sentiment Analysis Model on the cloud It goes over the building, hyperparameter tuning and deployment of a machine learning model using Amazon Sagemaker. This is the last topic in this nanodegree. Goes over code implementations of various different GAN architectures like Deep Convolutional GANs, Pix2Pix & CycleGAN. Brief introduction to Attention in context of deep learning and go over a basic code implementation.Ĭovers GANs in depth, taught by the inventor, Ian Goodfellow himself. Goes over hyperparameter tuning and embeddings in neural networks by implementing Word2Vec model. Goes over different weight initialization strategies for improved performance, an overview of style transfer, autoencoders and transfer learning.Įxplores how memory can be incorporated into a deep learning model using RNNs. Project: Bike sharing pattern prediction using multilayered perceptron.Ĭovers CNNs in depth alongside its different architectures, implementation and uses in various scenarios. Introduction to PyTorch, gradient descent algorithm, implementing and training a perceptron. TopicsĪ theoretical introduction to deep learning and a refresher on Anaconda, Numpy, matrix math and Jupyter Notebooks. Each topic ends with an open-ended project which has to be implemented in a Jupyter Notebook, either on your computer or their workspace and the files are then to be zipped and submitted for review.Īll the projects have to be marked complete by the reviewer in order to graduate. The machine learning framework used to code throughout the degree is PyTorch. The virtual machine runs an instance of Jupyter Notebook and can be accessed from the browser. Each topic consists of multiple video lessons and sometimes a hands-on lab on their virtual machine known as the the Udacity Workspace. A little programming experience in python and knowledge of basic linear algebra is recommended.Īs of May 2020, the Deep learning Nanodegree consists of 6 topics.People with a little prior experience and knowledge of ML.
People looking for a more code focused program rather than one which emphasizes on the math.People having no prior knowledge wanting to get started with deep learning from the basics.I recently graduated from the Udacity Deep Learning Nanodegree program and would like to provide a concise review of the degree.
0 kommentar(er)
