Deep Neural Networks for Supervised Learning: Regression
In Chapters 1 and 2 , we explored the topic of DL and studied how DL evolved from ML to solve an interesting area of problems. We discussed the need for DL frameworks and briefly explored a few popular frameworks available in the market. We then studied
- PDF / 733,133 Bytes
- 47 Pages / 439.37 x 666.142 pts Page_size
- 84 Downloads / 244 Views
Deep Neural Networks for Supervised Learning: Regression In Chapters 1 and 2, we explored the topic of DL and studied how DL evolved from ML to solve an interesting area of problems. We discussed the need for DL frameworks and briefly explored a few popular frameworks available in the market. We then studied why Keras is special and spent some time playing around with its basic building blocks provided to develop DNNs and also understood the intuition behind a DL model holistically. We then put together all our learnings from the practical exercises to develop a baby neural network for the Boston house prices use case. Now that we have a fair understanding of the different DL building blocks and the associated science, let’s explore a practical DNN for a regression use case in this chapter.
Getting Started The evolution of AI as a field and the increasing number of researchers and practitioners in the field have created a mature and benevolent community. Today, it’s fairly easy to access tools, research papers, datasets, and in fact even infrastructure to practice DL as a field. For our first use © Jojo Moolayil 2019 J. Moolayil, Learn Keras for Deep Neural Networks, https://doi.org/10.1007/978-1-4842-4240-7_3
53
Chapter 3
Deep Neural Networks for Supervised Learning: Regression
case, we would need a dataset and a business problem to get started. Here are a few popular choices. •
Kaggle: www.kaggle.com/ Kaggle is the world’s largest community of data scientists and machine learners. It started off as an online ML competition forum and later evolved into a mature platform that is highly recommended for every individual in data science. It still hosts ML competitions and also provides ML datasets, kernels or community-developed scripts for solving ML problems, ML jobs, and a platform to develop and execute ML models for the hosted competitions and public datasets.
•
US Government Open Data: www.data.gov/ Provides access to thousands of datasets on agriculture, climate, finance, and so on.
•
Indian Government Open Data: https://data.gov.in/ Provides open datasets for India’s demography, education, economy, industries, and so on.
•
Amazon Web Service Datasets: https://registry. opendata.aws/ Provides a few large datasets from NASA NEX and Openstreetmap, the Deutsche Bank public dataset, and so on.
•
54
Google Dataset Search: https://toolbox.google. com/datasetsearch
Chapter 3
Deep Neural Networks for Supervised Learning: Regression
This is relatively new and still in beta (at the writing of this book), but very promising. It provides access to thousands of public datasets for research experiments with a simple search query. It aggregates datasets from several public dataset repositories. •
UCI ML Repository: https://archive.ics.uci.edu/ml/ Another popular repository to explore datasets for ML and DL.
We will use the Kaggle public data repository for getting datasets for our DL use case. We will use the Rossmann Store sales dataset, which is available at www.kaggle.com/c/rossmann-store-sales
Data Loading...
