How-To Tutorials - Artificial Intelligence

I’d been preparing for Thanksgiving a good 3 weeks in advance. One reason is that I’d recently rented out a new apartment and the troops were heading over to my place this year. I obviously had to make sure everything went well and for that, trust me, there was no resting even for a minute! Thanksgiving is really about being thankful for the people and things in your life and spending quality time with family. This Thanksgiving I’m especially grateful to Amazon for making it the best experience ever! Read on to find out how Amazon made things awesome! Good times started two weeks ago when I was at the AmazonGo store with my friend, Sue. [embed]https://www.youtube.com/watch?v=NrmMk1Myrxc[/embed] In fact, this was the first time I had set foot in one of the stores. I wanted to see what was so cool about them and why everyone had been talking about them for so long! The store was pretty big and lived up to the A to Z concept, as far as I could see. The only odd thing was that I didn’t notice any queues or a billing counter. Sue glided around the floor with ease, as if she did this every day. I was more interested in seeing what was so special about this place. After she got her stuff, she headed straight for the door. I smiled to myself thinking how absent minded she was. So I called her back and reminded her “You haven’t gotten your products billed.” She smiled back at me and shrugged, “I don’t need to.” Before I could open my mouth to tell her off for stealing, she explained to me about the store. It’s something totally futuristic! Have you ever imagined not having to stand in a line to buy groceries? At the store, you just had to log in to your AmazonGo app on your phone, enter the store, grab your stuff and then leave. The sensors installed everywhere in the store automatically detected what you’d picked up and would bill you accordingly. They also used Computer Vision and Deep Learning to track people and their shopping carts. Now that’s something! And you even got a receipt! Well, it was my birthday last week and knowing what an avid reader I was, my colleagues from office gifted me a brand new Kindle. I loved every bit of it, but the best part was the X-ray feature. With X-ray, you could simply get information about a character, person or terms in a book. You could also scroll through the long lists of excerpts and click on one to go directly to that particular portion of the book! That’s really amazing, especially if you want to read a particular part of the book quickly. It came in use at the right time - I downloaded a load of recipe books for the turkey. Another feather in the cap for Amazon! Talking about feathers in one’s cap, you won’t believe it, but Amazon actually got me rekognised at work a few days ago. Nah, that wasn’t a typo. I worked as a software developer/ML engineer in a startup and I’d been doing this for as long as I can remember. I recently built this cool mobile application that recognized faces and unlocked your phone even when you didn’t have something like Face ID on your phone and the app had gotten us a million downloads in a month! It could also recognize and give you information about the ethnicity of a person if you captured their photograph with the phone’s camera. The trick was that I’d used the AmazonRekognition APIs for enhanced face detection in the application. Rekognition allows you to detect objects, scenes, text, and faces, using highly scalable, deep learning models. I also enhanced the application using the Polly API. Polly converts text to whichever language you want the speech in and gives you the synthesized speech in the form of audio files.The app I built now converted input text into 18 different languages, helping one converse with the person in front of them in that particular language, should they have a problem doing it in English. I got that long awaited promotion right after! Ever wondered how I got the new apartment? ;) Since the folks were coming over to my place in a few days, I thought I’d get a new dinner set. You’d probably think I would need to sit down at my PC or probably pick up my phone to search for a set online, but I had better things to do. Thanks to Alexa, I simply needed to ask her to find one for me and she did it brilliantly. Now Alexa isn’t my girlfriend, although I would have loved that to be. Alexa is actually Amazon’s cloud-based voice service that provides customers with an engaging way of interacting with technology. Alexa is blessed with finely tuned ASR or Automatic Speech Recognition and NLU or Natural Language Understanding engines, that instantly recognize and respond to voice requests. I selected a pretty looking set and instantly bought it through my Prime account. With technology like this at my fingertips, the developer in me had left no time in exploring possibilities with Alexa. That’s when I found out about Lex, built on the same deep learning platform that Alexa works on, which allows developers to build conversational interfaces into their apps. With the dinner set out of the way, I sat back with my feet up on the table. I was awesome, baby! Oh crap! I forgot to buy the turkey, the potatoes, the wine and a whole load of other stuff. It was 3 AM and I started panicking. I remembered that mum always put the turkey in the fridge at least 3 days in advance. I had only 2! I didn’t even have the time to make it to the AmazonGo store. I was panicking again and called up Suzy to ask her if she could pick up the stuff for me. She sounded so calm over the phone when I narrated my horror to her. She simply told me to get the stuff from AmazonFresh. So I hastily disconnected the call and almost screamed to Alexa, “Alexa, find me a big bird!”, and before I realized what I had said, I was presented with this. [caption id="attachment_2215" align="aligncenter" width="184"] Big Bird is one of the main protagonist in Sesame Street.[/caption] So I tried again, this time specifying what I actually needed! With AmazonDash integrating with AmazonFresh, I was able to get the turkey and other groceries delivered home in no time! What a blessing, indeed! A day before Thanksgiving, I was stuck in the office, working late on a new project. We usually tinkered around with a lot of ML and AI stuff. There was this project which needed the team to come up with a really innovative algorithm to perform a deep learning task. As the project lead, I was responsible for choosing the tech stack and I’m glad a little birdie had recently told me about AWS taking in MXNet officially as a Deep Learning Framework. MXNet made it a breeze to build ML applications that train quickly and could run anywhere. Moreover, with the recent collaboration between Amazon and Microsoft, a new ML library called Gluon was born. Available in MXNet, Gluon made building ML models, even more, easier and quicker, without compromising on performance. Need I say the project was successful? I got home that evening and sat down to pick a good flick or two to download from Amazon PrimeVideo. There’s always someone in the family who’d suggest we all watch a movie and I had to be prepared. With that done I quickly showered and got to bed. It was going to be a long day the next day! 4 AM my alarm rang and I was up! It was Thanksgiving, and what a wonderful day it was! I quickly got ready and prepared to start cooking. I got the bird out of the freezer and started to thaw it in cold water. It was a big bird so it was going to take some time. In the meantime, I cleaned up the house and then started working on the dressing. Apples, sausages, and cranberry. Yum! As I sliced up the sausages I realized that I had misjudged the quantity. I needed to get a couple more packets immediately! I had to run to the grocery store right away or there would be a disaster! But it took me a few minutes to remember it was Thanksgiving, one of the craziest days to get out on the road. I could call the store delivery guy or probably Amazon Dash, but then that would be illogical cos he’d have to take the same congested roads to get home.  I turned to Alexa for help, “Alexa, how do I get sausages delivered home in the next 30 minutes?”. And there I got my answer - Try Amazon PrimeAir. Now I don’t know about you, but having a drone deliver a couple packs of sausages to my house, is nothing less than ecstatic! I sat it out near the window for the next 20 minutes, praying that the package wouldn’t be intercepted by some hungry birds! I couldn’t miss the sight of the pork flying towards my apartment. With the dressing and turkey baked and ready, things were shaping up much better than I had expected. The folks started rolling in by lunchtime. Mum and dad were both quite impressed with the way I had organized things. I was beaming and in my mind hi-fived Amazon for helping me make everything possible with its amazing products and services designed to delight customers. It truly lives up to its slogan: Work hard. Have fun. Make history. If you are one of those folks who do this every day, behind the scenes, by building amazing products powered by machine learning and big data to make other's lives better, I want to thank you today for all your hard work. This Thanksgiving weekend, Packt's offering an unbelievable deal - Buy any book or video for just $10 or any three for $25! I know what I have my eyes on! Python Machine Learning - Second Edition by Sebastian Raschka and Vahid Mirjalili Effective Amazon Machine Learning by Alexis Perrier OpenCV 3 - Advanced Image Detection and Reconstruction [Video] by Prof. Robert Laganiere In the end, there’s nothing better than spending quality time with your family, enjoying a sumptuous meal, watching smiles all around and just being thankful for all you have. All I could say was, this Thanksgiving was truly Amazon fulfilled! :) Happy Thanksgiving folks!    

We often think of Data science and machine learning as skills essential to a niche group of researchers, data scientists, and developers. But the world as we know today revolves around data and algorithms, just as it used to revolve around programming a decade back. As data science and algorithms get integrated into all aspects of businesses across industries, data science like Microsoft Excel will become ubiquitous and will serve as a handy tool which makes you better at your job no matter what your job is. Knowing data science is key to having a bright career in this algoconomy (algorithm driven economy). If you are big on new year resolutions, make yourself a promise to carve your place in the algorithm-powered world by becoming data science savvy. Follow these three resolutions to set yourself up for a bright data-driven career. Get the foundations right: Start with the building blocks of data science, i.e. developing your technical skills. Stay relevant: Keep yourself updated on the latest developments in your field and periodically invest in reskilling and upskilling. Be mindful of your impact: Finally, always remember that your work has real-world implications. Choose your projects wisely and your project goals, hypotheses, and contributors with even more care. In this three-part series, we expand on how data professionals could go about achieving these three resolutions. But the principles behind the ideas are easily transferable to anyone in any job. Think of them as algorithms that can help you achieve your desired professional outcome! You simply need to engineer the features and fine-tune the hyperparameters specific to your industry and job role. 1st Resolution: Learn the building blocks of data science If you are interested in starting a career in data science or in one that involves data, here is a simple learning roadmap for you to develop your technical skills. Start off with learning a data-friendly programming language, one that you find easy and interesting. Next, brush up your statistics skills. Nothing fancy, just your high school math and stats would do nicely. Next, learn about algorithms - what they do, what questions they answer, how many types are there and how to write one. Finally, you can put all that learning to practice by building models on top of your choice of Machine Learning framework. Now let’s see, how you can accomplish each of these tasks 1. Learn Python or any another popular data friendly programming language you find interesting (Learning period: 1 week - 2 months) If you see yourself as a data scientist in the near future, knowing a programming language is one of the first things to check off your list. We suggest you learn a data-friendly programming language like Python or R. Python is a popular choice because of its strong, fast, and easy computational capabilities for the Data Science workflow. Moreover, because of a large and active community, the likelihood of finding someone in your team or your organization who knows Python is quite high, which is an added advantage. “Python has become the most popular programming language for data science because it allows us to forget about the tedious parts of programming and offers us an environment where we can quickly jot down our ideas and put concepts directly into action.” - Sebastian Raschka We suggest learning the basics from the book Learn Python in 7 days by Mohit, Bhaskar N. Das. Then you can move on to learning Python specifically for data science with Python Data Science Essentials by Alberto Boschetti. Additionally, you can learn R, which is a highly useful language when it comes to statistics and data. For learning R, we recommend R Data science Essentials by Raja B. Koushik. You can learn more about how Python and R stand against each other in the data science domain here. Although R and Python are the most popular choices for new developers and aspiring data scientists, you can also use Java for data science, if that is your cup of tea. Scala is another alternative. 2. Brush up on Statistics (Learning period: 1 week - 3 weeks) While you are training your programming muscle, we recommend that you brush through basic mathematics (probability and statistics). Remember, you already know everything to get started with data science from your high school days. You just need to refresh your memory with a little practice. A good place to start is to understand concepts like standard deviation, probability, mean, mode, variance, kurtosis among others. Now, your normal high-school books should be enough to get started, however, an in-depth understanding is required to leverage the power of data science. We recommend the book Statistics for Data Science by James D. Miller for this. 3. Learn what machine learning algorithms do and which ones to learn (Learning period: 1 month - 3 months) Machine Learning is a powerful tool to make predictions based on huge amounts of data. According to a recent study, in the next ten years, ML algorithms are expected to replace a quarter of the jobs across the world, in fields like transport, manufacturing, architecture, healthcare and many others. So the next step in your data science journey is learning about machine learning algorithms. There are new algorithms popping up almost every day. We’ve collated a list of top ten algorithms that you should learn to effectively design reliable and robust ML systems. But fear not, you don’t need to know all of them to get started. Start with some basic algorithms that are majorly used in the real world applications like linear regression, naive bayes, and decision trees. 4. Learn TensorFlow, Keras, or any other popular machine learning framework (Learning period: 1 month - 3 months) After you have familiarized yourself with some of the machine learning algorithms, it is time you put that learning to practice by building models based on those algorithms. While there are many cloud-based machine learning options that have click-based model building features available, the best way to learn a skill is to get your hands dirty. There is a growing range of frameworks that make it easy to build complex models while allowing for high degrees of customization. Here is a list of top 10 deep learning frameworks at your disposal to choose from. Our favorite pick is TensorFlow. It’s Python-based, backed by Google, has a very good documentation, and there are tons of tutorials and videos available on the internet to guide you. You can find a comprehensive list of books for learning Tensorflow here. We also recommend learning Keras, which is a good option if you have some knowledge of Python programming and want to get started with deep learning. Try the book Deep Learning with Keras, by Antonio Gulli and Sujit Pal, to get you started. If you find learning from multiple sources daunting, just learn from Sebastian Raschka’s Python machine learning book.   Once you have got your fundamentals right, it is important to stay relevant through continuous learning and reskilling. Check out part 2 where we explore how you could about doing this in a systematic and time efficient manner. In part 3, we look at ways you can own your work and become aware of its outcome.

In this article by Kishore Gaddam, author of the book Building Bots with Microsoft Bot Framework, we introduced what is Microsoft Bot Framework and how it helps in the development of bots. (For more resources related to this topic, see here.) Since past several decades, the corporate, government, and business world has experienced several waves of IT architecture foundations, moving from mainframes, to minicomputers, to distributed PCs, to the Internet, to social/mobile and now the Cloud/Internet of Thuings (IoT) Stack. We call this the Sixth wave of Corporate IT, and like its predecessors, Cloud and IoT technologies are causing significant disruption and displacement, even while it drives new levels of productivity. Each architecture focused on key business processes and supported killer technology applications to drive new levels of value. Very soon we will be looking at an enormous networked interconnection of everyday machines to one another, as well as to humans. Machine-to-machine-to-human connectivity will have a profound impact on the consumer and corporate IT experience. As these machines become social and talkto us, we have enormous opportunity to greatly enhance their value proposition through improved product quality, customer experience, and lowered cost of operations. A heightened consumer expectation for more personal and real-time interactions is driving business to holistically embrace the next wave of technology innovation like Cloud, IoT, and Bots to boost business performance. In this age of billions of connected devices, there is a need for such a technology where our apps could talk back, like bots? Bots that have specific purposes and talk to any device or any app or to anyone, Bots that live in cloud, Bots that we can talk to you via any communication channel such as email, text, voice, chat, and others. Bots can go where no apps have gone before when it comes to machine-to-machine-to-human connectivity. And to make this happen we will need a whole new platform. A Platform for Conversations. Conservation as a Service (CaaS) Messaging apps in general are becoming a second home screen for many people, acting as their entry point into the internet. And where the youngins are, the brands will follow. Companies are coming to messaging apps as bots and apps, to offer everything from customer service to online shopping and banking. Conversations are shaping up be the next major human-computer interface. Thanks to advances in natural language processing and machine learning, the tech is finally getting fast and accurate enough to be viable. Imagine a platform where language is the new UI layer. When we talk about conversation as a platform, there are 3 parts: There are people talking to people – Skype translator as an ex where people can communicate across cross languages Then there is the presence or being able to enhance a conversation by the ability to be present and interact remotely Then there is personal assistance and the bots Think of Bots as the new mechanism that you can converse with. Instead of looking through multiple mobile apps or pages and pages of websites, you can call on any application as a bot within the conversational canvas. Bots are the new apps and digital assistants are the meta apps. This way intelligence is infused into all our interactions. This leads us to Microsoft Bot Framework, which is a comprehensive offering from Microsoft to build and deploy high quality bots for your users to interact using Conversation as a Platform (CaaP). This is a framework that lets you build and connect intelligent bots. The idea is that they interact naturally wherever your users are talking, like Skype, Slack, Facebook Messenger, Text/SMS, and others. Basically any kind of channel that you use today as a human being to talk to other people, well, you will be able to use them to talk to bots all using natural language. Microsoft Bot Framework is a Microsoft operated CaaP service and an open source SDK. Bot Framework is one of the many tools Microsoft is offering to for building a complete Bot. Other tools include Language Understanding Intelligent Service (LUIS), Speech APIs, Microsoft Azure, Cortana Intelligence Suit and many more. Your Bot The Microsoft Bot Builder SDK is one of three main components of the Microsoft Bot Framework. First you have to build your bot. Your bot lives in the cloud and you host it yourself. You write it just like a web service component using Node.js or C#, like a ASP.NET WebAPI component. Microsoft Bot builder SDK is open source and so you will have more languages and web stack get supported over time. Your bot will have its own logic, but you also need a conversation logic using dialogs to model a conversation. The Bot builder SDK gives you facilities for this and there are many types of dialogs that are included from simple Yes/No questions to full natural language understanding or LUIS, which is one of the API's provided in Microsoft Cognitive Services: Bot Connector Bot Connector is hosted and operated by Microsoft. Think of it as a central router between your bots and many channels to communicate with your bots. Apart from routing messages, it would be managing state within the conversation. The Bot Connector is an easy way to create a single back-end and then publish to a bunch of different platforms called channels. Bot Directory Bot Directory is where user will be able to find bots. It's like app store for mobile apps. The Bot Directory is a public directory of all reviewed bots registered through the developer portal. Users will be able to discover, try, and add bots to their favorite conversation experiences from the Bot Directory. Anyone can access it and anyone can submit Bots to the directory. As you begin your development with Microsoft Bot Framework, you might be wondering how to best get started. Bots can be built in C#, however, Microsoft's Bot Framework can also be used to build bots using Node.js. For developing any bots, we need to first setup the development environment and have the right tools installed for successfully developing and deploying a bot. Let's see how we can setup a development environment using Visual Studio. Setting up development environment Let's first look at the Prerequisites required to set up the development environment: Prerequisites To use the Microsoft Bot Framework Connector, you must have: A Microsoft Account (Hotmail, Live, or Outlook) to log into the Bot Framework developer portal, which you will use to register your Bot. An Azure subscription (Free trial: https://azure.microsoft.com/en-us/). This Azure subscription is essential for having an Azure-accessible REST endpoint exposing a callback for the Connector service. Developer accounts on one or more communication services (such as Skype, Slack, Facebook) where your Bot will communicate. In addition, you may wish to have an Azure App Insights account so you can capture telemetry from your Bot. There are additionally different ways to go about building a Bot; from scratch, coded directly to the Bot Connector REST API, the Bot Builder SDK's for Node.js and .NET, and the Bot Connector .NET template which is what this quick start guide demonstrates. Setting up Bot Framework Connector SDK .NET This is a step-by-step guide to setting up dev environment to develop a Bot in C# using the Bot Framework Connector SDK .NET template: Install prerequisite software Visual Studio 2015 (latest update) - you can download the community version here for free: www.visualstudio.com Important: Please update all Visual Studio extensions to their latest versions to do so navigate to Tools | Extensions and Updates | Updates Download and install the Bot Application template: Download the file from the direct download link at http://aka.ms/bf-bc-vstemplate Save the zip file to your Visual Studio 2015 templates directory which is traditionally in %USERPROFILE%DocumentsVisual Studio 2015TemplatesProjectTemplatesVisual C# Open Visual Studio. Create a new C# project using the new Bot Application template. The template is a fully functional Echo Bot that takes the user's text utterance as input and returns it as output. In order to run however: The bot has to be registered with Bot Connector The AppId and AppPassword from the Bot Framework registration page have to be recorded in the project's web.config The project needs to be published to the web Emulator Use the Bot Framework Emulator to test your Bot application. The Bot Framework provides a channel emulator that lets you test calls to your Bot as if it were being called by the Bot Framework cloud service. To install the Bot Framework Emulator, download it from https://download.botframework.com/bf-v3/tools/emulator/publish.html. One installed, you're ready to test. First, start your Bot in Visual Studio using a browser as the application host. The following screenshot uses Microsoft Edge: Summary In this article, we introduced what is Microsoft Bot Framework and how it helps in the development of bots. Also, we have seen how to setup development environment, Emulator and the tools needed for programming. This article is based on the thought that programming knowledge and experience grow best when they grow together.  Resources for Article: Further resources on this subject: Talking to Bot using Browser [article] Webhooks in Slack [article] Creating our first bot, WebBot [article]

We have already talked about a simple learning roadmap for you to develop your data science skills in the first resolution. We also talked about the importance of staying relevant in an increasingly automated job market, in our second resolution. Now it’s time to think about the kind of person you want to be and the legacy you will leave behind. 3rd Resolution: Choose projects wisely and be mindful of their impact. Your work has real consequences. And your projects will often be larger than what you know or can do. As such, the first step toward creating impact with intention is to define the project scope, purpose, outcomes and assets clearly. The next most important factor is choosing the project team. 1. Seek out, learn from and work with a diverse group of people To become a successful data scientist you must learn how to collaborate. Not only does it make projects fun and efficient, but it also brings in diverse points of view and expertise from other disciplines. This is a great advantage for machine learning projects that attempt to solve complex real-world problems. You could benefit from working with other technical professionals like web developers, software programmers, data analysts, data administrators, game developers etc. Collaborating with such people will enhance your own domain knowledge and skills and also let you see your work from a broader technical perspective. Apart from the people involved in the core data and software domain, there are others who also have a primary stake in your project’s success. These include UX designers, people with humanities background if you are building a product intended to participate in society (which most products often are), business development folks, who actually sell your product and bring revenue, marketing people, who are responsible for bringing your product to a much wider audience to name a few. Working with people of diverse skill sets will help market your product right and make it useful and interpretable to the target audience. In addition to working with a melange of people with diverse skill sets and educational background it is also important to work with people who think differently from you, and who have experiences that are different from yours to get a more holistic idea of the problems your project is trying to tackle and to arrive at a richer and unique set of solutions to solve those problems. 2. Educate yourself on ethics for data science As an aspiring data scientist, you should always keep in mind the ethical aspects surrounding privacy, data sharing, and algorithmic decision-making.  Here are some ways to develop a mind inclined to designing ethically-sound data science projects and models. Listen to seminars and talks by experts and researchers in fairness, accountability, and transparency in machine learning systems. Our favorites include Kate Crawford’s talk on The trouble with bias, Tricia Wang on The human insights missing from big data and Ethics & Data Science by Jeff Hammerbacher. Follow top influencers on social media and catch up with their blogs and about their work regularly. Some of these researchers include Kate Crawford, Margaret Mitchell, Rich Caruana, Jake Metcalf, Michael Veale, and Kristian Lum among others. Take up courses which will guide you on how to eliminate unintended bias while designing data-driven algorithms. We recommend Data Science Ethics by the University of Michigan, available on edX. You can also take up a course on basic Philosophy from your choice of University.   Start at the beginning. Read books on ethics and philosophy when you get long weekends this year. You can begin with Aristotle's Nicomachean Ethics to understand the real meaning of ethics, a term Aristotle helped develop. We recommend browsing through The Stanford Encyclopedia of Philosophy, which is an online archive of peer-reviewed publication of original papers in philosophy, freely accessible to Internet users. You can also try Practical Ethics, a book by Peter Singer and The Elements of Moral Philosophy by James Rachels. Attend or follow upcoming conferences in the field of bringing transparency in socio-technical systems. For starters, FAT* (Conference on Fairness, Accountability, and Transparency) is scheduled on February 23 and 24th, 2018 at New York University, NYC. We also have the 5th annual conference of FAT/ML, later in the year.  3. Question/Reassess your hypotheses before, during and after actual implementation Finally, for any data science project, always reassess your hypotheses before, during, and after the actual implementation. Always ask yourself these questions after each of the above steps and compare them with the previous answers. What question are you asking? What is your project about? Whose needs is it addressing? Who could it adversely impact? What data are you using? Is the data-type suitable for your type of model? Is the data relevant and fresh? What are its inherent biases and limitations? How robust are your workarounds for them? What techniques are you going to try? What algorithms are you going to implement? What would be its complexity? Is it interpretable and transparent? How will you evaluate your methods and results? What do you expect the results to be? Are the results biased? Are they reproducible? These pointers will help you evaluate your project goals from a customer and business point of view. Additionally, it will also help you in building efficient models which can benefit the society and your organization at large. With this, we come to the end of our new year resolutions for an aspiring data scientist. However, the beauty of the ideas behind these resolutions is that they are easily transferable to anyone in any job. All you gotta do is get your foundations right, stay relevant, and be mindful of your impact. We hope this gives a great kick start to your career in 2018. “Motivation is what gets you started. Habit is what keeps you going.” ― Jim Ryun Happy New Year! May the odds and the God(s) be in your favor this year to help you build your resolutions into your daily routines and habits!

Generative Adversarial Models, introduced by Ian Goodfellow, are the next big revolution in the field of deep learning. Why? Because of their ability to perform semi-supervised learning where there is a vast majority of data is unlabelled. Here, GANs can efficiently carry out image generation tasks and other tasks such as converting sketches to an image, conversion of satellite images to a map, and many other tasks. GANs are capable of generating realistic images in any circumstances, for instance, giving some text written in a particular handwriting as an input to the generative model in order to generate more texts in the similar handwriting. The speciality of these GANs is that as compared to discriminative models, these generative models make use of a joint distribution probability to generate more likely samples. In short, these generative models or GANs are an improvisation to the discriminative models. Let’s explore some of the research papers that are contributing to further advancements in GANs. CycleGAN: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks This paper talks about CycleGANs, a class of generative Adversarial networks that carry out Image-to-Image translation. This means, capturing special characteristics of one image collection and figuring out how these characteristics could be translated into the other image collection, all in the absence of any paired training examples. CycleGANs method can also be applied in variety of applications such as Collection Style Transfer, Object Transfiguration, season transfer and photo enhancement. Cycle GAN architecture Source: GitHub CycleGANs are built upon the advantages of PIX2PIX architecture. The key advantage of CycleGANs model is, it allows to point the model at two discrete, unpaired collection of images.For example, one image collection say Group A, would consist photos of landscapes in summer whereas Group B would include photos of  landscapes in winter. The CycleGAN model can learn to translate the images between these two aesthetics without the need to merge tightly correlated matches together into a single X/Y training image. Source: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks The way CycleGANs are able to learn such great translations without having explicit X/Y training images involves introducing the idea of a full translation cycle to determine how good the entire translation system is, thus improving both generators at the same time. Source: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks Currently, the applications of CycleGANs can be seen in Image-to-Image translation and video translations. For example they can be seen used in Animal Transfiguration, Turning portrait faces into doll faces, and so on. Further ahead, we could potentially see its implementations in audio, text, etc., would help us in generating new data for training. Although this method has compelling results, it also has some limitations The geometric changes within an image are not fully successful (for instance, the cat to dog transformation showed minute success). This could be caused by the generator architecture choices, which are tailored for good performance on the appearance changes. Thus, handling more varied and extreme transformations, especially geometric changes, is an important problem. Failure caused by the distribution characteristics of the training datasets. For instance, in the horse to zebra transfiguration, the model got confused as it was trained on the wild horse and zebra synsets of ImageNet, which does not contain images of a person riding a horse or zebra. These and some other limitations are described in the research paper. To read more about CycleGANs in detail visit the link here. Wasserstein GAN In this paper, we get an exposure to Wasserstein GANs and how they overcomes the drawbacks in original GANs. Although GANs have shown a drastic success in realistic image generation, the training however is not that easy as the process is slow and unstable. In the paper proposed for WGANs, it is empirically shown that WGANs cure the training problem. Wasserstein distance, also known as Earth Mover’s (EM) distance, is a measure of distance between two probability distributions. The basic idea in WGAN is to replace the loss function so that there always exists a non-zero gradient. This can be done using Wasserstein distance between the generator distribution and the data distribution. Training these WGANs does not require keeping a balance in training of the discriminator and the generator. It also doesn’t require a design of the network architecture too. One of the most fascinating practical benefits of WGANs is the ability to continuously estimate the EM distance by training the discriminator to an optimal level. The learning curves when used for plotting are useful for debugging and hyperparameter searches. These curves also correlate well with the observed sample quality and improved stability of the optimization process. Thus, Wasserstein GANs are an alternative to traditional GAN training with features such as: Improvement in the stability of learning Elimination of problems like mode collapse Provide meaningful learning curves useful for debugging and hyperparameter searches Furthermore, the paper also showcases that the corresponding optimization problem is sound, and provides extensive theoretical work highlighting the deep connections to other distances between distributions. The Wasserstein GAN has been utilized to train a language translation machine. The condition here is that there is no parallel data between the word embeddings between the two languages. Wasserstein GANs have been used to perform English-Russian and English-Chinese language mappings. Limitations of WGANs: WGANs suffer from unstable training at times, when one uses a momentum based optimizer or when one uses high learning rates. Includes slow convergence after weight clipping, especially when clipping window is too large. It also suffers from the vanishing gradient problem when the clipping window is too small. To have a detailed understanding of WGANs have a look at the research paper here. InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets This paper describes InfoGAN (Information-theoretic extension to the Generative Adversarial Network). It can learn disentangled representations in a completely unsupervised manner. In traditional GANs, learned dataset is entangled i.e. encoded in a complex manner within the data space. However, if the representation is disentangled, it would be easy to implement and easy to apply tasks on it. InfoGAN solves the entangled data problem in GANs.   Specifically, InfoGAN successfully disentangles writing styles from digit shapes on the MNIST dataset, extracts poses of objects correctly irrespective of the lighting conditions within the 3D rendered images, and background digits from the central digit on the SVHN dataset. It also discovers visual concepts that include hairstyles, presence/absence of eyeglasses, and emotions on the CelebA face dataset. InfoGAN does not require any kind of supervision. In comparison to InfoGAN, the only other unsupervised method that learns disentangled representations is hossRBM, a higher-order extension of the spike-and-slab restricted Boltzmann machine which disentangles emotion from identity on the Toronto Face Dataset. However, hossRBM can only disentangle discrete latent factors, and its computation cost grows exponentially in the number of factors. Whereas, InfoGAN can disentangle both discrete and continuous latent factors, scale to complicated datasets, and typically requires no more training time than regular GAN. In the experiments given in the paper, firstly the comparison of  InfoGAN with prior approaches on relatively clean datasets is shown. Another experiment shown is, where InfoGAN can learn interpretable representations on complex datasets (here no previous unsupervised approach is known to learn representations of comparable quality.) Thus, InfoGAN is completely unsupervised and learns interpretable and disentangled representations on challenging datasets. Additionally, InfoGAN adds only negligible computation cost on top of GAN and is easy to train. The core idea of using mutual information to induce representation can be applied to other methods like VAE (Variational AutoEncoder) in future. The other possibilities with InfoGAN in future could be,learning hierarchical latent representations, improving semi-supervised learning with better codes, and using InfoGAN as a high-dimensional data discovery tool. To know more about this research paper in detail, visit the link given here. Progressive growing of GANs for improved Quality, Stability, and Variation This paper describes a brand new method for training your Generative Adversarial Networks. The basic idea here is to train both the generator and the discriminator progressively. This means, starting from a low resolution and adding new layers so that the model increases in providing images with finer details as training progresses. Such a method speeds up the training and also stabilizes it to a greater extent, which in turn produces images of unprecedented quality. For instance, a higher quality version of the CELEBA images dataset that provides output resolutions up to 10242 pixels.   Source: https://arxiv.org/pdf/1710.10196.pdf When new layers are added to the networks, they fade in smoothly. This helps in avoiding the sudden shocks to the already well-trained, smaller resolution layers. Also, the progressive training has various other benefits. The generation of smaller images is substantially more stable because there is less class information and fewer modes By increasing the resolution little by little, we are continuously asking a much simpler question compared to the end goal of discovering a mapping from latent vectors to e.g. 10242 images Progressive growing of GANs also reduces the training time. In addition to this, most of the iterations are done at lower resolutions, and the quality of the result obtained is upto 2-6 times faster, depending on the resolution of the final output. Thus, by progressively training GANs results into better quality, stability, and variation in images. This may also lead to true photorealism in near future. The paper concludes with the fact that, though there are certain limitations with this training method, which include semantic sensibility and understanding dataset-dependent constraints(such as certain objects being straight rather than curved). This leaves a lot to be desired from GANs and there is also room for improvement in the micro-structure of the images. To have a thorough understanding of this research paper, read the paper here.  

In this article by Eduardo Freitas, the author of the book Building Bots with Node.js, we will learn about Internet Relay Chat (IRC). It enables us to communicate in real time in the form of text. This chat runs on a TCP protocol in a client server model. IRC supports group messaging which is called as channels and also supports private message. (For more resources related to this topic, see here.) IRC is organized into many networks with different audiences. IRC being a client server, users need IRC clients to connect to IRC servers. IRC Client software comes as a packaged software as well as web based clients. Some browsers are also providing IRC clients as add-ons. Users can either install on their systems and then can be used to connect to IRC servers or networks. While connecting these IRC Servers, users will have to provide unique nick or nickname and choose existing channel for communication or users can start a new channel while connecting to these servers. In this article, we are going to develop one of such IRC bots for bug tracking purpose. This bug tracking bot will provide information about bugs as well as details about a particular bug. All this will be done seamlessly within IRC channels itself. It's going to be one window operations for a team when it comes to knowing about their bugs or defects. Great!! IRC Client and server As mentioned in introduction, to initiate an IRC communication, we need an IRC Client and Server or a Network to which our client will be connected. We will be using freenode network for our client to connect to. Freenode is the largest free and open source software focused IRC network. IRC Web-based Client I will be using IRC web based client using URL(https://webchat.freenode.net/). After opening the URL, you will see the following screen, As mentioned earlier, while connecting, we need to provide Nickname: and Channels:. I have provided Nickname: as Madan and at Channels: as #BugsChannel. In IRC, channels are always identified with #, so I provided # for my bugs channel. This is the new channel that we will be starting for communication. All the developers or team members can similarly provide their nicknames and this channel name to join for communication. Now let's ensure Humanity: by selecting I'm not a robot and click button Connect. Once connected, you will see the following screen. With this, our IRC client is connected to freenode network. You can also see username on right hand side as @Madan within this #BugsChannel. Whoever is joining this channel using this channel name and a network, will be shown on right hand side. In the next article, we will ask our bot to join this channel and the same network and will see how it appears within the channel. IRC bots IRC bot is a program which connects to IRC as one of the clients and appears as one of the users in IRC channels. These IRC bots are used for providing IRC Services or to host chat based custom implementations which will help teams for efficient collaboration. Creating our first IRC bot using IRC and NodeJS Let's start by creating a folder in our local drive in order to store our bot program from the command prompt. mkdir ircbot cd ircbot Assuming we have Node.js and NPM installed and let's create and initialize our package.json, which will store our bot's dependencies and definitions. npm init Once you go through the npm init options (which are very easy to follow), you'll see something similar to this. On your project folder you'll see the result which is your package.json file. Let's install irc package from NPM. This can be located at https://www.npmjs.com/package/irc. In order to install it, run this npm command. npm install –-save irc You should then see something similar to this. Having done this, the next thing to do is to update your package.json in order to include the "engines" attribute. Open with a text editor the package.json file and update it as follows. "engines": { "node": ">=5.6.0" } Your package.json should then look like this. Let's create our app.js file which will be the entry point to our bot as mentioned while setting up our node package. Our app.js should like this. var irc = require('irc'); var client = new irc.Client('irc.freenode.net', 'BugTrackerIRCBot', { autoConnect: false }); client.connect(5, function(serverReply) { console.log("Connected!n", serverReply); client.join('#BugsChannel', function(input) { console.log("Joined #BugsChannel"); client.say('#BugsChannel', "Hi, there. I am an IRC Bot which track bugs or defects for your team.n I can help you using following commands.n BUGREPORT n BUG # <BUG. NO>"); }); }); Now let's run our Node.js program and at first see how our console looks. If everything works well, our console should show our bot as connected to the required network and also joined a channel. Console can be seen as the following, Now if you look at our channel #BugsChannel in our web client, you should see our bot has joined and also sent a welcome message as well. Refer the following screen: If you look at the the preceding screen, our bot program got has executed successfully. Our bot BugTrackerIRCBot has joined the channel #BugsChannel and also bot sent an introduction message to all whoever is on channel. If you look at the right side of the screen under usernames, we are seeing BugTrackerIRCBot below @Madan Code understanding of our basic bot After seeing how our bot looks in IRC client, let's look at basic code implementation from app.js. We used irc library with the following lines, var irc = require('irc'); Using irc library, we instantiated client to connect one of the IRC networks using the following code snippet, var client = new irc.Client('irc.freenode.net', 'BugTrackerIRCBot', { autoConnect: false }); Here we connected to network irc.freenode.net and provided a nickname as BugTrackerIRCBot. This name has been given as I would like my bot to track and report the bugs in future. Now we ask client to connect and join a specific channel using the following code snippet, client.connect(5, function(serverReply) { console.log("Connected!n", serverReply); client.join('#BugsChannel', function(input) { console.log("Joined #BugsChannel"); client.say('#BugsChannel', "Hi, there. I am an IRC Bot which track bugs or defects for your team.n I can help you using following commands.n BUGREPORT n BUG # <BUG. NO>"); }); }); In preceeding code snippet, once client is connected, we get reply from server. This reply we are showing on a console. Once successfully connected, we ask bot to join a channel using the following code lines: client.join('#BugsChannel', function(input) { Remember, #BugsChannel is where we have joined from web client at the start. Now using client.join(), I am asking my bot to join the same channel. Once bot is joined, bot is saying a welcome message in the same channel using function client.say(). Hope this has given some basic understanding of our bot and it's code implementations. In the next article, we will enhance our bot so that our teams can have effective communication experience while chatting itself. Enhancing our BugTrackerIRCBot Having built a very basic IRC bot, let's enhance our BugTrackerIRCBot. As developers, we always would like to know how our programs or a system is functioning. To do this typically our testing teams carry out testing of a system or a program and log their bugs or defects into a bug tracking software or a system. We developers later can take a look at those bugs and address them as a part of our development life cycle. During this journey, developers will collaborate and communicate over messaging platforms like IRC. We would like to provide unique experience during their development by leveraging IRC bots. So here is what exactly we are doing. We are creating a channel for communication all the team members will be joined and our bot will also be there. In this channel, bugs will be reported and communicated based on developers' request. Also if developers need some additional information about a bug, chat bot can also help them by providing a URL from the bug tracking system. Awesome!! But before going in to details, let me summarize using the following steps about how we are going to do this, Enhance our basic bot program for more conversational experience Bug tracking system or bug storage where bugs will be stored and tracked for developers Here we mentioned about bug storage system. In this article, I would like to explain DocumentDB which is a NoSQL JSON based cloud storage system. What is DocumentDB? I have already explained NoSQLs. DocumentDB is also one of such NoSQLs where data is stored in JSON documents and offered by Microsoft Azure platform. Details of DocumentDB can be referred from (https://azure.microsoft.com/en-in/services/documentdb/) Setting up a DocumentDB for our BugTrackerIRCBot Assuming you already have a Microsoft Azure subscription follow these steps to configure DocumentDB for your bot. Create account ID for DocumentDB Let's create a new account called botdb using the following screenshot from Azure portal. Select NoSQL API as of DocumentDB. Select appropriate subscription and resources. I am using existing resources for this account. You can also create a new dedicated resource for this account. Once you enter all the required information, hit Create button at the bottom to create new account for DocumentDB. Newly created account botdb can be seen as the following, Create collection and database Select a botdb account from account lists shown precedingly. This will show various menu options like Properties, Settings, Collections etc. Under this account we need to create a collection to store bugs data. To create a new collection, click on Add Collection option as shown in the following screenshot, On click of Add Collection option, following screen will be shown on right side of the screen. Please enter the details as shown in the following screenshot: In the preceding screen, we are creating a new database along with our new collection Bugs. This new database will be named as BugDB. Once this database is created, we can add other bugs related collections in future in the same database. This can be done in future using option Use existing from the preceding screen. Once you enter all the relevant data, click OK to create database as well as collection. Refer the following screenshot: From the preceding screen, COLLECTION ID and DATABASE shown will be used during enhancing our bot. Create data for our BugTrackerIRCBot Now we have BugsDB with Bugs collection which will hold all the data for bugs. Let's add some data into our collection. To add a data let's use menu option Document Explorer shown in the following screenshot: This will open up a screen showing list of Databases and Collections created so far. Select our database as BugDB and collection as Bugs from the available list. Refer the following screenshot: To create a JSON document for our Bugs collection, click on Create option. This will open up a New Document screen to enter JSON based data. Please enter a data as per the following screenshot: We will be storing id, status, title, description, priority,assignedto, url attributes for our single bug document which will get stored in Bugs collection. To save JOSN document in our collection click Save button. Refer the following screenshot: This way we can create sample records in bugs collection which will be later wired up in NodeJS program. Sample list of bugs can be seen in the following screenshot: Summary Every development team needs bug tracking and reporting tools. There are typical needs of bug reporting and bug assignment. In case of critical projects these needs become also very critical for project timelines. This article showed us how we can provide a seamless experience to developers while they are communicating with peers within a channel. To summarize so far, we understood how to use DocumentDB from Microsoft Azure. Using DocumentDB, we created a new collection along with new database to store bugs data. We also added some sample JSON documents in Bugs collection. In today's world of collaboration, development teams who would be using such integrations and automations would be efficient and effective while delivering their quality products. Resources for Article: Further resources on this subject: Talking to Bot using Browser [article] Asynchronous Control Flow Patterns with ES2015 and beyond [article] Basic Website using Node.js and MySQL database [article]

The Canadian Parliament's Standing Committee on Access to Information, Privacy and Ethics hosted the hearing of the International Grand Committee on Big Data, Privacy and Democracy from Monday May 27 to Wednesday May 29.  Witnesses from at least 11 countries appeared before representatives to testify on how governments can protect democracy and citizen rights in the age of big data. This section of the hearing, which took place on May 28, includes Roger McNamee’s take on why Silicon Valley wants to build data voodoo dolls for users. Roger McNamee is the Author of Zucked: Waking up to the Facebook Catastrophe. His remarks in this section of the hearing builds on previous hearing presentations by Professor Zuboff, Professor Park Ben Scott and the previous talk by Jim Balsillie. Roger McNamee’s remarks build on previous hearing presentations by Professor Zuboff, Professor Park Ben Scott and the previous talk by Jim Balsillie. He started off by saying, “Beginning in 2004, I noticed a transformation in the culture of Silicon Valley and over the course of a decade customer focused models were replaced by the relentless pursuit of global scale, monopoly, and massive wealth.” McNamee says that Google wants to make the world more efficient, they want to eliminate user stress that results from too many choices. Now, Google knew that society would not permit a business model based on denying consumer choice and free will, so they covered their tracks. Beginning around 2012, Facebook adopted a similar strategy later followed by Amazon, Microsoft, and others. For Google and Facebook, the business is behavioral prediction using which they build a high-resolution data avatar of every consumer--a voodoo doll if you will. They gather a tiny amount of data from user posts and queries; but the vast majority of their data comes from surveillance, web tracking, scanning emails and documents, data from apps and third parties, and ambient surveillance from products like Alexa, Google assistant, sidewalk labs, and Pokemon go. Google and Facebook used data voodoo dolls to provide their customers who are marketers with perfect information about every consumer. They use the same data to manipulate consumer choices just as in China behavioral manipulation is the goal. The algorithms of Google and Facebook are tuned to keep users on site and active; preferably by pressing emotional buttons that reveal each user's true self. For most users, this means content that provokes fear or outrage. Hate speech, disinformation, and conspiracy theories are catnip for these algorithms. The design of these platforms treats all content precisely the same whether it be hard news from a reliable site, a warning about an emergency, or a conspiracy theory. The platforms make no judgments, users choose aided by algorithms that reinforce past behavior. The result is, 2.5 billion Truman shows on Facebook each a unique world with its own facts. In the U.S. nearly 40% of the population identifies with at least one thing that is demonstrably false; this undermines democracy. “The people at Google and Facebook are not evil they are the products of an American business culture with few rules where misbehavior seldom results in punishment”, he says. Unlike industrial businesses, internet platforms are highly adaptable and this is the challenge. If you take away one opportunity they will move on to the next one and they are moving upmarket getting rid of the middlemen. Today, they apply behavioral prediction to advertising but they have already set their sights on transportation and financial services. This is not an argument against undermining their advertising business but rather a warning that it may be a Pyrrhic victory. If a user’s goals are to protect democracy and personal liberty, McNamee tells them, they have to be bold. They have to force a radical transformation of the business model of internet platforms. That would mean, at a minimum banning web tracking, scanning of email and documents, third party commerce and data, and ambient surveillance. A second option would be to tax micro targeted advertising to make it economically unattractive. But you also need to create space for alternative business models using trust that longs last. Startups can happen anywhere they can come from each of your countries. At the end of the day, though the most effective path to reform would be to shut down the platforms at least temporarily as Sri Lanka did. Any country can go first. The platform's have left you no choice the time has come to call their bluff companies with responsible business models will emerge overnight to fill the void. McNamee explains, “when they (organizations) gather all of this data the purpose of it is to create a high resolution avatar of each and every human being. Doesn't matter whether they use their systems or not they collect it on absolutely everybody. In the Caribbean, Voodoo was essentially this notion that you create a doll, an avatar, such that you can poke it with a pin and the person would experience that pain right and so it becomes literally a representation of the human being.” To know more you can listen to the full hearing video titled, “Meeting No. 152 ETHI - Standing Committee on Access to Information, Privacy and Ethics” on ParlVU. Experts present most pressing issues facing global lawmakers on citizens’ privacy, democracy and rights to freedom of speech Time for data privacy: DuckDuckGo CEO Gabe Weinberg in an interview with Kara Swisher Over 19 years of ANU(Australian National University) students’ and staff data breached

Lise Getoor is a professor in the Computer Science Department, at the University of California, Santa Cruz. She has a PhD in Computer Science from Stanford University. She has spent a lot of time studying machine learning, reasoning under uncertainty, databases, data science for social good, artificial intelligence This article attempts to bring our readers to Lisa’s Keynote speech at NIPS 2017. It highlights how structures can be unreasonably effective and the ways to leverage structures in Machine learning problems. After reading this article, head over to the NIPS Facebook page for the complete keynote. All images in this article come from Lisa’s presentation slides and do not belong to us. Our ability to collect, manipulate, analyze, and act on vast amounts of data is having a profound impact on all aspects of society. Much of this data is heterogeneous in nature and interlinked in a myriad of complex ways. This Data is Multimodal (it has different kinds of entities), Multi-relational (it has different links between things), and Spatio-Temporal (it involves space and time parameters). This keynote explores how we can exploit the structure that's in the input as well as the output of machine learning algorithms. A large number of structured problems exists in the fields of NLP and computer vision, computational biology, computational social sciences, knowledge graph extraction and so on. According to Dan Roth, all interesting decisions are structured i.e. there are dependencies between the predictions. Most ML algorithms take this nicely structured data and flatten it put it in a matrix form, which is convenient for our algorithms. However, there is a bunch of different issues with it. The most fundamental issue with the matrix form is that it assumes incorrect independence. Further, in the context of structure and outputs, we’re unable to apply the collective reasoning about the predictions we made for different entries in this matrix. Therefore we need to have ways where we can declaratively talk about how to transform the structure into features. This talk provides us with patterns, tools, and templates for dealing with structures in both inputs and outputs.   Lisa has covered three topics for solving structured problems: Patterns, Tools, and Templates. Patterns are used for simple structured problems. Tools help in getting patterns to work and in creating tractable structured problems. Templates build on patterns and tools to solve bigger computational problems. [dropcap]1[/dropcap] Patterns They are used for naively simple structured problems. But on encoding them repeatedly, one can increase performance by 5 or 10%. We use Logical Rules to capture structure in patterns. These logical structures capture structure, i.e. they give an easy way of talking about entities and links between entities. They also tend to be interpretable. There are three basic patterns for structured prediction problems: Collective Classification, Link Prediction, Entity Resolution. [toggle title="To learn more about Patterns, open this section" state="close"] Collective Classification Collective classification is used for inferring the labels of nodes in a graph. The pattern for expressing this in logical rules is [box type="success" align="" class="" width=""]local - predictor (x, l) → label (x, l) label (x, l) & link(x,y) → label (y,l)[/box] It is called as collective classification as the thing to predict i.e. the label, occurs on both sides of the rule. Let us consider a toy problem: We have to predict unknown labels here (marked in grey) as to what political party the unknown person will vote for. We apply logical rules to the problem. Local rules: [box type="success" align="" class="" width=""]“If X donates to part P, X votes for P” “If X tweets party P slogans, X votes for P”[/box] Relational rules: [box type="success" align="" class="" width=""]“If X is linked to Y, and X votes for P, Y votes for P” Votes (X,P) & Friends (X,Y) → Votes (Y, P) Votes (X,P) & Spouse (X,Y) → Votes (Y, P)[/box] The above example shows the local and relational rules applied to the problem based on collective classification. Adding a collective classifier like this to other problems yields significant improvement. Link Prediction Link Prediction is used for predicting links or edges in a graph. The pattern for expressing this in logical rules is : [box type="success" align="" class="" width=""]link (x,y) & similar (y,z) →  link (x,y)[/box] For example, consider a basic recommendation system. We apply logical rules of link prediction to express likes and similarities. So, how you infer one link is gonna give you information about another link. Rules express: [box type="success" align="" class="" width=""]“If user U likes item1, and item2 is similar to item1, user U likes item2” Likes (U, I1) & SimilarItem (I1, I2) → Likes(U, I2) “If user1  likes item I, and user2 is similar to user1, user2 likes item I” Likes (U1, I) & SimilarUser (U1, U2) → Likes(U2, I)[/box] Entity Resolution Entity Resolution is used for determining which nodes refer to the same underlying entity. Here we use local rules between how similar things are, for instance, how similar their names or links are [box type="success" align="" class="" width=""]similar - name (x,y) → same (x,y) similar - links (x,y) → same (x,y)[/box] There are two collective rules. One is based on transitivity. [box type="success" align="" class="" width=""]similar - name (x,y) → same (x,y) similar - links (x,y) → same (x,y) same (x,y) && same(y,z) → same (x,z)[/box] The other is based on matching i.e. dependence on both sides of the rule. [box type="success" align="" class="" width=""]similar - name (x,y) → same (x,y) similar - links (x,y) → same (x,y) same (x,y) & ! same (y,z) → ! same (x,z)[/box] The logical rules as described above though being quite helpful, have certain disadvantages. They are intractable, can’t handle inconsistencies, and can’t represent degrees of similarity.[/toggle] [dropcap]2[/dropcap] Tools Tools help in making the structured kind of problems tractable and in getting patterns to work. The tools come from the Statistical Relational Learning community.  Lise adds another one to this mix of languages - PSL. PSL is probabilistic logical programming, a declarative language for expressing collective inference problems. To know more: psl.linqs.org Predicate = relationship or property Ground Atom = (continuous) random variable Weighted Rules = capture dependency or constraint PSL Program = Rules + Input DB PSL makes reasoning scalable by mapping Logical inference to Convex optimization. The language takes logical rules and assign weights to them and then uses it to define a distribution for the unknown variables. One of the striking features here is that the random variables have continuous values. The work done pertaining to the PSL language turns the disadvantages of logical rules into advantages. So they are tractable, can handle inconsistencies, and can represent similarity. The key idea is to convert the clauses to concave functions. To be tractable, we relax it to a concave maximization. PSL has semantics from three different worlds: Randomized algorithms from the Computer science community, Probabilistic graphical models from the Machine Learning community, and Soft Logic from the AI community. [toggle title="To learn more about PSL, open this section" state="close"] Randomized Algorithm In this setting, we have a weighted rule. We have nonnegative weights and then a set of weighted logical rules in clausal form. Weighted Max SAT is a classical problem where we attempt to find the assignment to the random variables that maximize the weights of the satisfied rules. However, this problem is NP-HARD (which is a computational complexity theory for non-deterministic polynomial-time hardness). To overcome this, the randomized community converts this combinatorial optimization to a continuous optimization by introducing random variables which denote rounding probabilities. Probabilistic Graphic Models Graph models represent problems as a factor graph where we have random variables and rules that are essentially the potential function. However, this problem is also NP-Hard. We use Variational Inference approximation technique to solve this. Here we introduce marginal distributions (μ) for the variables. We can then express a solution if we can find a set of globally consistent assignment for these marginal distributions. The problem here is, although we can express it as a linear program, there is an exponential number of constraints. We will use techniques from the graphical model's community, particularly Local Consistency Relaxation to convert this to a simpler problem. The simple idea is to relax search over consistent marginals to simpler set. We introduce local pseudo marginals over joint potential states. Using KKT conditions we can optimize out the θ to derive simplified projected LCR over μ. This approach shows 16% improvement over canonical dual decomposition (MPLP) Soft Logic In the Soft Logic technique for convex optimizations, we have random variables that denote a degree of truth or similarity. We are essentially trying to minimize the amount of dissatisfaction in the rules. Hence with three different interpretations i.e. Randomized Algorithms, Graphical Models, and Soft Logic, we get the same convex optimizations. A PSL essentially takes a PSL program, takes some input data and defines a convex optimization. PSL is open-source. The code, data, tutorials are available online at psl.linqs.org MAP inference in PSL translates into convex optimization problem Inference is further enhanced with state-of-the-art optimization and distributed graph processing paradigms Learning methods for rule weights and latent variables Using PSL gives fast as well as accurate results on comparison with other approaches. [/toggle] [dropcap]3[/dropcap] Templates Templates build on patterns to solve problems in bigger areas such as computational social sciences, knowledge discovery, and responsible data science and machine learning. [toggle title="To learn about some use cases of PSL and Templates for pattern recognition, open this section." state="close"] Computational Social Sciences For exploring this area we will apply a PSL model to Debate stance classification. Let us consider a scenario of an online debate. The topic of the debate is climate change. We can use information in the text to figure out if the people participating in the debate are pro or anti the topic. We can also use information about the dialogue in the discourse. And we can build this on a PSL model. This is based on the collective classification problem we saw earlier in the post. We get a significant rise in accuracy by using a PSL program. Here are the results Knowledge Discovery Using a structure and making use of patterns in Knowledge discovery really pays off. Although we have Information extractors which can extract information from the web and other sources such as facts about entities, relationships, they are usually noisy. So it gets difficult to reason about them collectively to figure out which facts we actually wanna add to our knowledge base. We can add structure to the knowledge graph construction by Performing collective classification, link prediction, and entity resolution Enforcing ontological constraints Integrate knowledge source confidences Using PSL to make it scalable Here’s the PSL program for knowledge graph identification. These were evaluated on three real-world knowledge graphs. NELL, MusicBrainz, and Freebase. As shown in the above image, both statistical features and semantic constraints help but combining them always wins. Responsible Machine Learning Understanding structure can be key to mitigating negative effects and lead to responsible Machine Learning. The perils of ignoring structure in the machine learning space include overlooking Privacy. For instance, many approaches consider only individual’s attribute data. Some don't take into account what can be inferred from relational context. The other area is around Fairness. The structure here is often outside the data. It can be in the organization or the socio-economic structure. To enable fairness we need to implement impartial decision making without bias and need to take into account structural patterns. Algorithmic Discrimination is another area which can make use of a structure. The fundamental structural pattern here is a feedback loop. Having a way of encoding this feedback loop is important to eliminate algorithmic discrimination. [/toggle] Conclusion In this article, we saw ways of exploiting structures that can be tractable. It provided some tools and templates for exploiting structure. The keynote also provided opportunities for Machine Learning methods that can mix: Structured and unstructured approaches Probabilistic and logical inference Data-driven and knowledge-driven modeling AI and Machine Learning developers need to build on the approaches as described above and discover, exploit, and find new structure and create compelling commercial, scientific, and societal applications.

A lot, I mean lots and lots of great articles have been written already about AI’s epic journey in 2017. They all generally agree that 2017 sets the stage for AI in very real terms.  We saw immense progress in academia, research and industry in terms of an explosion of new ideas (like capsNets), questioning of established ideas (like backprop, AI black boxes), new methods (Alpha Zero’s self-learning), tools (PyTorch, Gluon, AWS SageMaker), and hardware (quantum computers, AI chips). New and existing players gearing up to tap into this phenomena even as they struggled to tap into the limited talent pool at various conferences and other community hangouts. While we have accelerated the pace of testing and deploying some of those ideas in the real world with self-driving cars, in media & entertainment, among others, progress in building a supportive and sustainable ecosystem has been slow. We also saw conversations on AI ethics, transparency, interpretability, fairness, go mainstream alongside broader contexts such as national policies, corporate cultural reformation setting the tone of those conversations. While anxiety over losing jobs to robots keeps reaching new heights proportional to the cryptocurrency hype, we saw humanoids gain citizenship, residency and even talk about contesting in an election! It has been nothing short of the stuff, legendary tales are made of: struggle, confusion, magic, awe, love, fear, disgust, inspiring heroes, powerful villains, misunderstood monsters, inner demons and guardian angels. And stories worth telling must have songs written about them! Here’s our ode to AI Highlights in 2017 while paying homage to an all-time favorite: ‘A few of my favorite things’ from Sound of Music. Next year, our AI friends will probably join us behind the scenes in the making of another homage to the extraordinary advances in data science, machine learning, and AI. [box type="shadow" align="" class="" width=""] Stripes on horses and horsetails on zebras Bright funny faces in bowls full of rameN Brown furry bears rolled into pandAs These are a few of my favorite thinGs   TensorFlow projects and crisp algo models Libratus’ poker faces, AlphaGo Zero’s gaming caboodles Cars that drive and drones that fly with the moon on their wings These are a few of my favorite things   Interpreting AI black boxes, using Python hashes Kaggle frenemies and the ones from ML MOOC classes R white spaces that melt into strings These are a few of my favorite things   When models don’t converge, and networks just forget When I am sad I simply remember my favorite things And then I don’t feel so bad[/box]   PS: We had to leave out many other significant developments in the above cover as we are limited in our creative repertoire. We invite you to join in and help us write an extended version together! The idea is to make learning about data science easy, accessible, fun and memorable!