Explore
For Enterprise
Join for Free
Log In

Artes e Humanas
Negócios
Ciência da Computação
Ciência de Dados
Tecnologia da informação
da saúdesaúde
Matemática e Lógica
Desenvolvimento Pessoal
Ciência e Engenharia Física
Ciências Sociais
Línguas
Notas
Certificados

Explore todo o Coursera

Navegar
Pesquisar
For Enterprise
Entrar
Inscreva-se gratuitamente

Deep neural network training, tuning and prediction

Loading...

Sequences, Time Series and Prediction

deeplearning.ai

4.6 (228 classificações) | 5.9K alunos inscritos

Curso 4 de 4 no TensorFlow in Practice Programa de cursos integrados

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 Specialization will teach you best practices for using TensorFlow, a popular open-source framework for machine learning. In this fourth course, you will learn how to build time series models in TensorFlow. You’ll first implement best practices to prepare time series data. You’ll also explore how RNNs and 1D ConvNets can be used for prediction. Finally, you’ll apply everything you’ve learned throughout the Specialization to build a sunspot prediction model using real-world data! 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.

Visualizar o programa do curso

Habilidades que você aprenderá

Forecasting, Machine Learning, Tensorflow, Time Series, prediction

Avaliações

4.6 (228 classificações)

5 stars
169 ratings
4 stars
39 ratings
3 stars
10 ratings
2 stars
2 ratings
1 star
8 ratings

GG

Aug 05, 2019

Fantastic course, starting from basic fundamentals of statistical forecasting to using Convolutional neural networks. I will use my learnings directly to my job.

SD

Jul 31, 2019

Quite a good light-weighted course on Time Series and Prediction. It was quite helpful for people like me who are seeking ways to implement the concepts.

Na lição

Deep Neural Networks for Time Series

Having explored time series and some of the common attributes of time series such as trend and seasonality, and then having used statistical methods for projection, let's now begin to teach neural networks to recognize and predict on time series!

A conversation with Andrew Ng1:17

Preparing features and labels4:22

Feeding windowed dataset into neural network2:14

Single layer neural network2:53

Machine learning on time windows0:37

More on single layer neural network2:10

Deep neural network training, tuning and prediction4:18

Deep neural network3:03

Ministrado por

Laurence Moroney
AI Advocate

Transcrição

Okay, so this week you've been looking at machine learning for

predicting the next value in a sequence.

You've learned how to break your data down into window chunks that you could train

on, and then you saw a simple single layer neural network

that gave you what was effectively a linear regression.

Now let's take that to the next step with a DNN to see if we can improve

our model accuracy.

It's not that much different from the linear regression model we saw earlier.

And this is a relatively simple deep neural network that has three layers.

So let's unpack it line by line.

First we'll have to get a data set which will generate by passing in x_train data,

along with the desired window size, batch size, and shuffle buffer size.

We'll then define the model.

Let's keep it simple with three layers of 10, 10, and 1 neurons.

The input shape is the size of the window and

we'll activate each layer using a relu.

We'll then compile the model as before with a mean squared error loss function

and stochastic gradient descent optimizer.

Finally, we'll fit the model over 100 epochs, and

after a few seconds of training, we'll see results that look like this.

It's pretty good still.

And when we calculate the mean absolute error, we're lower than we were earlier,

so it's a step in the right direction.

But it's also a somewhat a stab in the dark,

particularly with the optimizer function.

Wouldn't it be nice if we could pick the optimal learning rate instead of the one

that we chose?

We might learn more efficiently and build a better model.

Now let's look at a technique for

that that uses callbacks that you used way back in the first course.

So here's a code for the previous neural network.

But I've added a callback to tweak the learning rate

using a learning rate scheduler.

You can see that code here.

This will be called at the callback at the end of each epoch.

What it will do is change the learning rates to a value

based on the epoch number.

So in epoch 1, it is 1 times 10 to the -8 times 10 to the power of 1 over 20.

And by the time we reach the 100 epoch,

it'll be 1 times 10 to the -8 times 10 to the power of 5, and that's 100 over 20.

This will happen on each callback because we set it

in the callbacks parameter of modeled outfit.

After training with this, we can then plot the last per epoch against

the learning rate per epoch by using this code, and we'll see a chart like this.

The y-axis shows us the loss for that epoch and

the x-axis shows us the learning rate.

We can then try to pick the lowest point of the curve where it's still relatively

stable like this, and that's right around 7 times 10 to the -6.

So let's set that to be our learning rate and then we'll retrain.

So here's the same neural network code, and

we've updated the learning rate, so we'll also train it for a bit longer.

Let's check the results after training for 500 epochs.

Here's the codes to plot out the loss that was calculated during the training, and

it will give us a chart like this.

Which upon first inspection looks like we're probably wasting our time

training beyond maybe only 10 epochs, but

it's somewhat skewed by the fact that the earlier losses were so high.

If we cropped them off and plot the loss for epochs after number 10 with code like

this, then the chart will tell us a different story.

We can see that the loss was continuing to decrease even after 500 epochs.

And that shows that our network is learning very well indeed.

And the results of the predictions overlaid against the originals looks

like this.

And the mean absolute error across the results

is significantly lower than earlier.

I'll take you through a screencast of this code in action in the next video.

Using a very simple DNN, we've improved our results very nicely.

But it's still just a DNN, there's no sequencing taken into account, and

in a time series like this, the values that are immediately before a value

are more likely to impact it than those further in the past.

And that's the perfect set up to use RNS like we had in the natural language

course.

Now we'll look at that next week, but first, let's dig into this code.

Explore nosso catálogo

Registre-se gratuitamente e obtenha recomendações, atualizações e ofertas personalizadas.

O Coursera proporciona acesso universal à melhor educação do mundo fazendo parcerias com as melhores universidades e organizações para oferecer cursos on-line.

© 2019 Coursera Inc. Todos os direitos reservados.

Baixar na App Store

Baixar no Google Play

Coursera
Sobre
Liderança
Carreiras
Lista de cursos
Certificados
Certificados MasterTrack™
Notas
For Enterprise
For Government

Comunidade
Learners
Partners
Developers
Beta Testers
Translators

Conectar
Blog
Facebook
LinkedIn
Twitter
Instagram
YouTube
Tech Blog

Mais
Termos
Privacidade
Ajuda
Acessibilidade
Imprensa
Contato
Diretório
Afiliados