{"id":8111,"date":"2020-08-27T09:00:00","date_gmt":"2020-08-27T07:00:00","guid":{"rendered":"https:\/\/www.codemotion.com\/magazine\/?p=8111"},"modified":"2022-01-05T20:06:00","modified_gmt":"2022-01-05T19:06:00","slug":"google-ai-hub","status":"publish","type":"post","link":"https:\/\/www.codemotion.com\/magazine\/ai-ml\/google-ai-hub\/","title":{"rendered":"Google AI Hub: what, why, how"},"content":{"rendered":"\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t
\n\t\t\t\t\t\t\tTable Of Contents\t\t\t\t\t\t<\/div>\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t
\n\t\t\t\t\t\t
  1. Why you may need artificial intelligence?<\/a>
  2. What is Google AI Hub?<\/a>
  3. How to benefit from Google AI Hub<\/a><\/ol>\t\t\t\t\t<\/div>\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\n\n\n

    Artificial intelligence<\/span> (AI<\/span>) and machine learning<\/span> (ML) increasingly seem to be indispensable tools<\/span> that developers<\/span> need to be able to handle. There are many ways these tools<\/span> can be put to use, applied to applications<\/span> and products<\/span>. In research<\/span> and academia, the subject has been around for 70 years or so – more or less the same time span which separates the birth of computers<\/span> and information technology<\/span> from the present day. However the popularity of this field has fluctuated considerably in the last few decades, experiencing dark times (the infamous \u2018AI<\/span> Winter\u2019) and golden eras, such as the present (a phase that does not seem destined to end any time soon).<\/p>\n\n\n\n

    Why you may need artificial intelligence?<\/h2>\n\n\n\n

    The immediate impact on everyday lives of Artificial Intelligence<\/span> and similar technologies<\/span> has never been as popular and widely (if not wildly) acknowledged as in the present day. Every CEO<\/span> wants their company<\/span> to use it, produce it, develop it – and every developer<\/span> wants to join the party. Of course, there is nothing wrong with that: on the contrary, for an entrepreneur it is a natural impulse to exploit state of the art technologies in order to keep pace with competitors and to try to take a forward step before them. It is also perfectly natural for a developer<\/span> to be intrigued, at the very least, by an impressive and pervasive technology<\/span> that, although still rather intricate from the theoretical point of view, is largely accessible in terms of both tools<\/span> and programming systems.<\/p>\n\n\n\n

    Even if you don\u2019t want to learn Python<\/span>, R<\/span> or Scala<\/span> (though you should!) and prefer to stick to the Java<\/span> and C#<\/span> you probably use in your daily work, ready to use libraries and frameworks will be found within your favourite computer language. If readers will permit a personal digression, my first experiences<\/span> with AI<\/span> were in BASIC(!) and my first professional project<\/span> in the field (being paid to deliver an AI product) some twenty years ago was in C: at the time I had to do most of the work \u2018by hand\u2019, due to a lack of standardised libraries (or indeed any libraries at all) suited to my purpose.<\/p>\n\n\n\n

    Today, things are simpler for developers<\/span> in this respect: one can learn a library or framework for an already-familiar language, or learn the foundations of an easier interactive<\/span> language<\/span>, such as Python or R, and start using de facto<\/em> the standard libraries such as TensorFlow<\/span> that are available for many mainstream languages (even for Javascript<\/span>).<\/p>\n\n\n\n

    In short, it is a natural and healthy instinct for a developer<\/span> to be interested in participating in and delivering AI projects. The easiest introduction involves finding tutorials<\/span>, explanations, or introductions written by other developers<\/span>, and downloading open source<\/span> tools<\/span>. Such tools<\/span> (Jupyter notebooks, for example) are usually easy to install and easy to use for those who are just starting to code and to solve problems using AI methods<\/span>.<\/p>\n\n\n\n

    Of course, where both CEOs and developers<\/span> (whose salaries<\/span> are paid by CEOs) want to work with AI<\/span>, it is obvious that the team<\/span>\u2019s joint efforts will result in the delivery of AI products<\/span> or solutions to sell to customers<\/span>.<\/p>\n\n\n\n

    However, it is precisely at this point that things become difficult: while a single developer<\/span> may create a Jupyter notebook that brilliantly solves some regression, prediction or generation problem, to transform that solitary effort into a standard delivery pipeline is very difficult \u2013 often, it may be better to restart from scratch.<\/p>\n\n\n\n

    On the one hand, projects<\/span> – collective efforts performed by teams<\/span> \u2013 are what leads to delivery; on the other hand, an enterprising solution needs to satisfy business requirements<\/span> – the first goal of any profitable project<\/span>. In other words, first the business<\/span> case, next the technology<\/span> required<\/span> to efficiently satisfy that need.<\/p>\n\n\n\n

    Developers<\/span> playing with Pytorch<\/span> late at night may produce interesting prototypes<\/span>, which may suggest ways to solve a problem or need experienced by the company<\/span> but creating a new product<\/span> on the strength of that idea alone is another matter entirely. A production pipeline with delivery of an AI-based product<\/span>, made for a specific purpose as its goal is needed, and will need to be managed properly. Artificial Intelligence<\/span> project<\/span> management is another interesting issue but will be dealt with elsewhere.<\/p>\n\n\n\n

    What is Google AI Hub?<\/h2>\n\n\n\n

    The time has now come to introduce our main character, Google<\/span> AI Hub<\/strong>: at first glance, this is just a repository<\/span> of tools<\/span> able to provide the individual parts of the pipeline mentioned above. It is also an ecosystem of plugins<\/span> and goes as far as supplying end-to-end pipelines to support the delivery of an AI<\/span> product<\/span>, at different levels of abstractions, according to the resources<\/span> available to produce it.<\/p>\n\n\n\n

    In fact, AI<\/span> Hub is more than a repository<\/span>, providing different assets for different goals: for example, it can be used to learn ML algorithms, or to use built artefacts available either in the public domain via Google<\/span>, or shared as plug-ins<\/span> within your organisation. Alternatively, one can use AI<\/span> Hub to share one\u2019s own models<\/span>, code and more with peers in the same organisation – a hub that facilitates collaboration<\/span> on AI<\/span> projects by means of reuse, sharing and factoring.Let\u2019s begin by finding something useful just to play with – something ready to use. Visit the site homepage<\/a> on which assets are classified in categories in a menu of the left hand side. Choose the \u2018Notebook\u2019 category for this example:<\/p>\n\n\n\n

    \"Google<\/figure><\/div>\n\n\n\n

    This offers a list of notebooks provided by Google<\/span>. For our current purposes, we could open the first and start using it.<\/p>\n\n\n\n

    \"\"<\/figure><\/div>\n\n\n\n

    Once we access the asset – in this case a Notebook – we can open it in Colab<\/span> to explore and exploit. This is a simple asset exploitation of course, but Google<\/span>-provided notebooks are great; well documented and easy to use, they\u2019re a good way to learn by doing.<\/p>\n\n\n\n

    Among the available assets we find datasets<\/span>, services (API<\/span>, for example, which may be called on by your application<\/span> to use built-in functionalities, or to train your model<\/span> via transfer learning, etc.), trained models<\/span>, TensorFlow modules<\/span>, virtual machine images, and Kubeflow pipelines. All these assets occur somewhere in the development process<\/span> of an AI application<\/span>. The importance of Kubeflow pipelines – an interesting way to embed AI models inside an application<\/span> \u2013 should be particularly stressed, but more on that later.<\/p>\n\n\n\n

    How to benefit from Google AI Hub<\/h2>\n\n\n\n

    In this introductory note we cannot give a general overview of all the tools<\/span> available on the Google<\/span> AI Hub dashboard (the platform itself provides several tutorials<\/span> on how to start using each tool<\/span> and resource<\/span> it makes available). In place of this, we offer some hints on the task of deploying a scalable<\/span> ML application<\/span> through the hub.<\/p>\n\n\n\n

    An important initial note about using AI Hub for practice is that you will need a Google Cloud<\/span> Platform account. Starter accounts that are essentially free of charge<\/span> are available, but you\u2019ll need to provide bank account<\/span> details. It\u2019s probably best to operate inside an organisation account instead – typically one belonging to your company<\/span>: organisations have the ability to use and share assets via the Hub. For example, if you work in R&D<\/span> you can share prototypes<\/span> with your colleagues working on architecture, delivery or another aspect of the product<\/span>.<\/p>\n\n\n\n

    \"Google<\/figure><\/div>\n\n\n\n

    The dashboard of the platform allows management<\/span> of projects<\/span> using assets from the hub. A project<\/span> may start as a simple Jupyter notebook<\/span>, for which you can choose not only the language (Python 2\/3, R, \u2026) but also the computational sizing (e.g. if you need some kind of GPU to properly run it, etc.) and other parameters. All of these factors determine the cost of the service needed to run the notebook.<\/p>\n\n\n\n

    \"\"<\/figure><\/div>\n\n\n\n
    \"\"<\/figure><\/div>\n\n\n\n

    Needless to say, you can edit and run your notebook on the cloud platform as you would in your local environment: you\u2019ll find all the main tools already available for whichever language and framework you chose; for example, TensorFlow<\/a> is already installed in the Python environments, and you can \u2018Pip\u2019 whatever additional packages you need.<\/p>\n\n\n\n

    \"Create<\/figure><\/div>\n\n\n\n

    It is also easy to pull and push your notebooks from and to Git repositories, or to containerize your notebook in order to install specific libraries and acquire the level of customization your code requires to run properly.<\/p>\n\n\n\n

    At a certain point (probably at the start!) you\u2019ll need to handle a dataset, perhaps to train your model or to fine tune a pre-trained model. AI Hub provides a section on datasets that is not simply a bookmark or repository but allows for labelling data. This is a practical need in many projects, and the lack of a dataset appropriate for your supervised model is a frequent issue when trying to build a product based on ML models.<\/p>\n\n\n\n

    \"Datasets<\/figure><\/div>\n\n\n\n

    In this section of the hub you can add a dataset for which you can specify the kind of data and its source, upload data and specify a label set which provides the complete list (to the best of your knowledge) of labels of your data. This is not only for recording purposes: in fact you can also add a set of instructions and rules according to which human labellers may attach labels to the elements of your dataset. This feature allows you to specify the requirements of a labelling activity to be performed by someone paid to do it on your behalf.<\/p>\n\n\n\n

    \"\"<\/figure><\/div>\n\n\n\n

    However, labelling data is not an easy task and is subject to ambiguities (people do this task instead of a machine for some very good reasons!) so one may need to refine instructions and initially provide a limited trial dataset on which to assess both the quality of labelling and the level of description actually required in the instructions. Since this is a crucial step in training a ML model, real life projects will require people to manage this activity by collaborating closely with the developers to get a useful, and as unbiased as possible, dataset on which to train the ML model.<\/p>\n\n\n\n

    \"\"<\/figure><\/div>\n\n\n\n
    \"\"<\/figure><\/div>\n\n\n\n

    \u2018Jobs\u2019 is another interesting feature from the AI platform. Used to train models, you may define these using standard built-in algorithms or your own algorithm, according to your model\u2019s needs. In most cases algorithms built in the platform will suffice for training purposes.<\/p>\n\n\n\n

    \"\"<\/figure><\/div>\n\n\n\n
    \"\"<\/figure><\/div>\n\n\n\n

    Up to this point we have talked about models, datasets (and the interesting labelling feature) and training jobs: these tasks form the bulk of an AI developer\u2019s day-to-day work, whether on their local systems or on the shared tools provided by their organisations.<\/p>\n\n\n\n

    A complete, end-to-end ML pipeline is somewhat more complicated, however, requiring at least the following steps<\/p>\n\n\n\n