To summarize this course then, we talked about the aim of the specialization, which is to teach you how to build production machine learning models whether you're a Python programmer or a data scientist. This specialization consists of a series of courses that provide a practical real-world introduction to machine learning. And then, we looked at how Google views machine learning as a way to replace heuristic rules that tend to build up over time. For example, Google search used heuristic rules to decide whether we needed to return a page for the San Francisco Giants or New York Giants. This is now done using machine learning. We then talked about machine learning as a way to increasingly personalize your business's offerings for your customers, using Google Maps as an example. For example, while routing directions between point A and point B can be done with a deterministic algorithm, inferring which floor of a subway station you're in needs some amount of machine learning. But building personalized recommendations, that's impossible to do at scale without ML. We then delved into the secret sauce behind machine learning. And it turns out to be something quite unglamorous. It's all about building organizational know-how. It's important to recognize the five phases that business processes tend to go through and not skip any of these stages in order to be successful at machine learning. It's also important to recognize how to transition between these phases thoughtfully. And finally, we talked about why fair is not the default in machine learning and how to recognize biases that can be amplified because of the data you use. Unconscious bias affects the way we do machine learning because it can get reinforced in the training data that we collect, it affects the way we collect and classify data, how we design and write code. To give you a taste of machine learning, we introduced Qwiklabs as a way to start trying out Google Cloud Platform. This is a way for you to do the labs in this specialization without having to pay extra for the computing resources. Qwiklabs also has a number of other labs and quests. I strongly encourage you to practice and gain experience in machine learning and Cloud technologies using Qwiklabs. We also told you how to find the source code. It's on GitHub, it's open source, and you should totally use our examples as a starting point for your projects. We also looked at Cloud Datalab. Python notebooks are the tool of choice for data scientists and is a way that you will do the majority of hands-on activities in this specialization. Now, the notebooks in this specialization are already worked out. So I encourage you to pause and think as you go through the notebooks step by step. And if you are up to the challenge, you could also remove some of the key cells and see if you can write the necessary code yourself. We talked about how Cloud Storage and Compute Engine provide the CPU and storage necessary for ephemeral distributed notebooks. And then, we used the notebook to launch BigQuery queries on thousands of machines at scale. This, for example, is a query that's carried out on a dataset of millions of healthcare claims. We also invoked pre-trained machine learning models that are available as APIs. As machine learning matures, many of the reusable tasks will be available in pre-trained form, whether it's Vision or Speech. A key point that these Machine Learning APIs teach us is that we want to take our Machine Learning models and make them just as easy to use. In this case, all of the Machine Learning APIs were REST APIs, they are microservices and they provided a very high level of abstraction. And with that, we come to the end of the first course of this specialization. Join us for the next course which will be on creating machine learning datasets.
