Quick TensorFlow's video lessons, practical projects, Colab notebooks and supplementary materials are designed to get you started building and deploying machine learning models in the real world as quickly as possible.
We'll guide you through an end-to-end ML workflow:
TensorFlow Datasets is a collection of datasets ready to use with the TensorFlow machine learning framework.
We'll teach you best-practices for building performant data pipelines for your machine learning models.
With Transfer Learning, you can use the "knowledge" from existing pre-trained models to empower your own custom models.
This course includes an in-depth discussion of various CNN architectures that you can use as a "base" for your models, including: MobileNet, EfficientNet, ResNet, and Inception
We then demonstrate how you can acess these models through both the Keras API and TensorFlow Hub.
Discover essential techniques that ensure your classification models perform well in production, starting with the best way to train-test split datasets and create validation subsets to avoid overfitting.
Learn how to effectively apply optimization strategies like hyper-parameter tuning and fine-tuning to improve training and validation accuracy, while comparing various optimization algorithms including SGD and Adam.
When you create an image classification model, the first thing you want to do is try it in the real world! But integrating machine learning models into usable applications can be challenging.
This course gives you early access to a new tool called PalletML , that lets you to deploy your image classification model to an Android app without any extra code.
Just upload your TensorFlow Lite model + labels, and Pallet takes care of hosting and serving your model through a mobile app. If you host your models on Google Drive, you can even deploy them right from your phone.
Configure real-time preprocessing options for images before model inference for improved accuracy: resizing, inversion, and grayscaling operations are available out-of-the-box.
Save predictions for later review, and filter through them using a custom interface designed for evaluating classification results.
Easily send your model to any number of friends and colleagues from within the Pallet app. A unique link is generated for each model that you deploy and opens directly to your model when activated.
The ML community has deployed numerous classifiers - from flower recognizers to clothing detectors - all direcly accessible in the Pallet app.
Nice to meet you and hope you enjoy the course!
Dr. Teate is a tenured Professor at James Madison University , and Director of the Data Science and Applied Machine Learning Laboratory.
His current areas of research include development and applications of machine learning models for the design of nanoscale semiconducting materials; small scale autonomous vehicles; image recognition; edge computing; and deep learning.