Ngesa Marvin - Keras (3) for the Curious and Creative | PyData Global 2023

Python

Learn how to harness the power of Keras 3 for deep learning, image classification, and more in this talk, covering APIs, activation functions, computer vision, and more.

Key takeaways

Keras is a Python library for deep learning, and Keras 3 is its new version.
Keras is more than just an API, it’s a deep learning framework that allows you to easily build models.
Keras has modules like keras.layers and keras.models for building and training models.
Keras also has features like activation functions, convolutional layers, pooling layers, and more.
The foundation of a neural network is a sequence of layers, and each layer has a specific function.
The layers are stacked on top of each other to create a deep network.
Keras allows you to create a deep learning model using a functional API or a sequential API.
TensorFlow and Keras work together seamlessly, and Keras also supports other backends like JAX.
Automatic differentiation is a feature of Keras that allows you to compute gradients and optimize models.
Keras has a lot of built-in support for computer vision, including image classification, object detection, and more.
The output of a neural network is a function of the input, and Keras makes it easy to define and train these functions.
Keras has many built-in optimization algorithms, including stochastic gradient descent, and also supports custom optimization algorithms.
Keras has many use cases, including image classification, object detection, natural language processing, and more.
The speaker also talked about a concept called super-resolution, which is a technique for generating high-resolution images from low-resolution ones.
Keras also supports text-to-image generation, and is capable of generating artistic images based on prompts.
The speaker promised to show more examples of how to use Keras in future talks.

Ngesa Marvin - Keras (3) for the Curious and Creative | PyData Global 2023

More talks