Antonio is currently at the Office of CTO at Google Cloud, where he leads innovation at the intersection of infrastructure and machine learning. He is also the author of Deep Learning with TensorFlow and Keras 2nd edition.
Deep Learning with Keras Antonio 27
[Antonio]: During the late 90s, I was attending university in Pisa and got interested in neural networks, optimization, and parallel computation. At that time, it was probably not considered the smartest choice because neural networks were kind of dying and going through the "AI winter," a period of reduced funding and interest in artificial intelligence research. Parallel computation was considered very hard because it needed to run on very expensive computers. In 1995, I had a chance to play with the open-source code of Lycos, which was at that time a research project and later a web search engine. That was truly fascinating. So, I decided to work in that direction. Since the very early days, web search had a lot of data science and machine learning. In 1996, I formed a company in Italy and we worked on one of the first public search engines ever, Arianna. Google was founded in 1998. Today, Arianna is a piece of software mentioned in the Computer History Museum . My passion for data science is rooted there and has never stopped.
In 2014, I read a few scientific papers on neural networks and started to think, "Hey, this thing seems to be having a comeback." The problem was finding a good software framework in which to develop machine learning models. Theano was a good one , but maybe a bit too low-level. I remember that Keras was impressive because it had the right level of abstraction for describing complex neural network models in a very simple and effective way. That was a perfect combination: neural networks were back and there was a good framework to play with.
To my knowledge, no one had written a book about the progress made in deep learning with Keras during the previous 3-4 years. That's the reason why in 2016, my friend Sujit and I decided to write Deep Learning with Keras. I was in a coffee shop in Amsterdam and Sujit was in California.
Keras is a beautiful API with the right level of abstraction; it is really well done. TensorFlow is an ecosystem with many different components. I got excited to see things like TPU support, federated learning, TensorFlow Graphics, and TensorFlow Agents. I also like the fact that you have many pre-trained models that can be used for things like transfer learning. Then there is eager execution, which has opened the door to interesting discussions in the community.
Do you think it closed the gap with PyTorch?I don't think there is a gap. Both the frameworks have done excellent work. TensorFlow is a large open-source machine learning ecosystem delivered by a team of AI experts and used in several thousand AI projects.
For instance, TensorFlow Extended (TFX) helps with building end-to-end machine learning pipelines. Then there is tooling to help model builders and model deployers understand their performance. A lot of effort has been put into integrations with beloved open-source tools like Numpy, SciPy, and Spacy.
There are two key points. First, we all want to live in a world where rather mundane tasks are avoided. For instance, what about a tool for finding a good combination of hyper-parameters on your behalf? And what about a tool that would automatically select the best deep learning model for my problem?
Personally, I believe that cloud and machine learning are offering companies the opportunity to provide innovative solutions for a very large number of customers. This transformation is about becoming a platform with unprecedented user bases and offering functionality as a service in a way that better satisfies the customer.
I now work for an organization called Office of the CTO in Google Cloud and my role is at the intersection between infrastructure and machine learning. I love to co-innovate with cloud partners and, together, solve challenging problems.
Deep Learning with TensorFlow 2 and Keras, Second Edition teaches neural networks and deep learning techniques alongside TensorFlow (TF) and Keras. You'll learn how to write deep learning applications in the most powerful, popular, and scalable machine learning stack available.
This book is for Python developers and data scientists who want to build machine learning and deep learning systems with TensorFlow. Whether or not you have done machine learning before, this book gives you the theory and practice required to use Keras, TensorFlow 2, and AutoML to build machine learning systems.
Educational Data Mining (EDM) is a research field that focuses on the application of data mining, machine learning, and statistical methods to detect patterns in large collections of educational data. Different machine learning techniques have been applied in this field over the years, but it has been recently that Deep Learning has gained increasing attention in the educational domain. Deep Learning is a machine learning method based on neural network architectures with multiple layers of processing units, which has been successfully applied to a broad set of problems in the areas of image recognition and natural language processing. This paper surveys the research carried out in Deep Learning techniques applied to EDM, from its origins to the present day. The main goals of this study are to identify the EDM tasks that have benefited from Deep Learning and those that are pending to be explored, to describe the main datasets used, to provide an overview of the key concepts, main architectures, and configurations of Deep Learning and its applications to EDM, and to discuss current state-of-the-art and future directions on this area of research.
The research field of Educational Data Mining (EDM) focuses on the application of techniques and methods of data mining in educational environments. EDM is concerned with developing, researching, and applying machine learning, data mining, and statistical methods to detect patterns in large collections of educational data that would otherwise be impossible to analyze [1]. 2ff7e9595c
Comments