Walking through training the ConvNet with fit_generator

Loading...

deeplearning.ai의 강좌에서

Introduction to TensorFlow for Artificial Intelligence, Machine Learning, and Deep Learning

1155개의 평가

deeplearning.ai

Introduction to TensorFlow for Artificial Intelligence, Machine Learning, and Deep Learning

1155개의 평가

If you are a software developer who wants to build scalable AI-powered algorithms, you need to understand how to use the tools to build them. This course is part of the upcoming Machine Learning in Tensorflow Specialization and will teach you best practices for using TensorFlow, a popular open-source framework for machine learning. The Machine Learning course and Deep Learning Specialization from Andrew Ng teach the most important and foundational principles of Machine Learning and Deep Learning. This new deeplearning.ai TensorFlow Specialization teaches you how to use TensorFlow to implement those principles so that you can start building and applying scalable models to real-world problems. To develop a deeper understanding of how neural networks work, we recommend that you take the Deep Learning Specialization. The full deeplearning.ai TensorFlow Specialization will be available later this year, but you can get started with Course 1, Introduction to Tensorflow for AI, ML and DL, available now on Coursera.

수업에서

Using Real-world Images

Last week you saw how to improve the results from your deep neural network using convolutions. It was a good start, but the data you used was very basic. What happens when your images are larger, or if the features aren’t always in the same place? Andrew and Laurence discuss this to prepare you for what you’ll learn this week: handling complex images!

A conversation with Andrew Ng2:20

Understanding ImageGenerator4:15

Defining a ConvNet to use complex images2:36

Training the ConvNet with fit_generator2:55

Walking through developing a ConvNet2:09

Walking through training the ConvNet with fit_generator3:29

Adding automatic validation to test accuracy4:44

Exploring the impact of compressing images3:09

강사 만나기

Laurence Moroney
AI Advocate
Google Brain

Next up is where we use the ImageDataGenerator.

We instantiate it and we scale our images to 1 over 255,

which then normalizes their values.

We then point it at the main directory where we see the unzipped files.

We can see that it finds all of the images, and

has assigned them to two classes, because they were two sub directories.

We'll now train the neural network for 15 epochs, it will take about two minutes.

Each epoch is loading the data, calculating the convolutions and

then trying to match the convolutions to labels.

As you can see, the accuracy mostly increases but it will occasionally deep,

showing the gradient ascent of the learning actually in action.

It's always a good idea to keep an eye on fluctuations in this figure.

And if there are too wild, you can adjust the learning rate.

Remember the parameter to RMS prop when you compile the model,

that's where you'd tweak it.

It's also going pretty fast, because right here, I'm training on a GPU machine.

By the time we get to epoch 15,

we can see that our accuracy is about 0.9981, which is really good.

But remember, that's only based on the data that the network has already

seen during training, which is only about 1,000 images.

So don't get lulled into a false sense of security.

Let's have a bit of fun with the model now and

see if we can predict the class for new images that it hasn't previously seen.

Let's go to Pixabay, and see what we can find.

I'll search for horses, and there's lots of horses, so let's pick this one.

It's a white horse running in the snow.

I'm going to download it to my file system.

I'm now going to go back to the notebook, and

I'm going to upload the image from my file system.

And we'll see that it gets uploaded, and it's classified as a horse.

So let's try another one.

Like this one here.

Which I'll then upload to the notebook, and

we'll see that it's also classified as a horse.

I'll now go back to Pixabay and search for person, and

pick this image of a girl sitting on a bench.

I'll download it to my file system, upload it to the neural network,

and we can see that this is also correctly classified as a human.

Let's do one more.

I'll go back to the list of results on Pixabay, and pick this image of a girl.

As before, I'll download it to my file system and I'll upload it to the neural

network and we'll also see that it's still detects a human in the image.

Now one other thing that I can do with this script is upload multiple files and

have it classify all of them at once.

And here we can see all of the classifications.

We have four out four correct.

This notebook also includes some visualizations of the image

as it passes through the convolutions.

You can give it a try with this script.

Here you can see where a human image was convolved and

features such as the legs really lit up.

And if I run it again, we can see another human with similar features.

Also the hair is very distinctive.

Have a play with it for yourself and see what you discover.

So there, we saw a convolutional neural network create a classifier to horses or

humans using a set of about 1,000 images.

The four images we tested all worked, but that's not really scalable.

And the next video, we'll see how we can add a validation set to the training and

have it automatically measure the accuracy of the validation set, too.

Coursera 카탈로그 살펴보기

무료로 참여해 맞춤화된 추천, 업데이트 및 제안을 받아보세요.

Coursera는 세계 최고의 교육에 대한 보편적인 접근 기회를 제공하며, 최고의 대학 및 기관들과 제휴하여 온라인 강좌를 제공합니다.

© 2019 Coursera Inc. All rights reserved.

앱스토어에서 다운로드하기

구글 플레이에서 받기