WebProNews

Category: MachineLearningPro

MachineLearningPro

  • How LinkedIn is Using Machine Learning to Determine Skills

    How LinkedIn is Using Machine Learning to Determine Skills

    One of the more interesting reveals that Dan Francis, Senior Product Manager for LinkedIn Talent Insights, provided in a recent talk about the Talent Insights tool is how LinkedIn is using machine learning to determine skills of people. He says that there are now over 575 million members in the LinkedIn database and there are 35,000 standardized skills in LinkedIn’s skills taxonomy. The way LinkedIn is figuring out what skills a member has is via machine learning technology.

    Dan Francis, Senior Product Manager, LinkedIn Talent Insights, discussed Talent Insights in a recent LinkedIn video embedded below:

    LinkedIn Using Machine Learning to Determine Skills

    The skills data in Talent Insights comes from a variety of sources, mainly from a member’s profile. There are over 35,000 standardized skills that we have in LinkedIn’s skills taxonomy, and the way we’re figuring out what skills a member has is using machine learning. We can identify skills that a member has that’s based on things that they explicitly added to their profile.

    The other thing that we’ll do is look at the text of the profile. There’s a field of machine learning called natural language processing and we’re basically using that. It’s scanning through all the words that are on a member’s profile, and when we can determine that it’s pertaining to the member, as oppose the company or another subject, we’ll say okay, we think that this member has this skill. We also look at other attributes, like their title or the company, to make sure they actually are very likely to have that skill.

    The last thing that we’ll do is look at the skills a member has and figure out what are skill relationships. So as an example, let’s say that a member has Ember, which is a type of JavaScript framework, since we know that they know Ember, they also know JavaScript. So if somebody’s running a search like that, we’ll surface them in the results. I think that the most important reason why this is helpful and the real benefit to users of the platform is when you’re searching, you want to get as accurate a view of the population as possible. What we’re trying to do is look at all the different signals that we possibly have to represent that view.  

    575 Million People on LinkedIn Globally and Adding 2 Per Second

    Today, LinkedIn has over 575 million members that are on the platform globally. This is actually growing at a pretty rapid clip, so we’re adding about two members per second. One of the great things about LinkedIn is that we’re actually very well represented in terms of the professional workforce globally. If you look at the top 30 economies around the world, we actually have the majority of professionals in all of those economies.

    LinkedIn is the World’s Largest Aggregator of Jobs

    I think there’s often a perception that most of the data’s directly from LinkedIn, stuff that’s posted on LinkedIn and job status is one notable exception to that. Plenty of companies and people will post jobs on LinkedIn, and that’s information that does get surfaced. However, we’re also the world’s largest aggregator of jobs. At this point there are over 20 million jobs that are on LinkedIn.

    The way that we’re getting that information is we’re working with over 40,000 partners. These are job boards, ATS’s, and direct customer relationships. We’re collecting all of those jobs, standardizing them, and showing them on our platform. The benefit is not just for displaying the data in Talent Insights, the benefit is also when members are searching on LinkedIn.com, we’re giving them as representative a view of the job market as possible.

  • AWS CEO Announces Textract to Extract Data Without Machine Learning Skills

    AWS CEO Announces Textract to Extract Data Without Machine Learning Skills

    AWS CEO Andy Jassy announced Amazon Textract at the AWS re:Invent 2018 conference. Textract allows AWS customers to automatically extract formatted data from documents without losing the structure of the data. Best of all, there are no machine learning skills required to use Textract. It’s something that many data-intensive enterprises have been requesting for many years.

    Amazon Launches Textract to Easily Extract Usable Data

    Our customers are frustrated that they can’t get more of all those text and data that are in documents into the cloud, so they can actually do machine learning on top of it. So we worked with our customers, we thought about what might solve these problems and I’m excited to announce the launch of Amazon Textract. This is an OCR plus plus service to easily extract text and data from virtually any document and there is no machine learning experience required.

    This is important, you don’t need to have any machine learning experience to be able to use Textract. Here’s how it generally works. Below is a pretty typical document, it’s got a couple of columns and it’s got a table in the middle of the left column.

    When you use OCR it just basically captures all that information in a row and so what you end up with is the gobbledygook you see in the box below which is completely useless. That’s typically what happens.

    Let’s go through what Textract does. Textract is intelligent. Textract is able to tell that there are two columns here so actually when you get the data and the language it reads like it’s supposed to be read. Textract is able to identify that there’s a table there and is able to lay out for you what that table should look like so you can actually read and use that data in whatever you’re trying to do on the analytics and machine learning side. That’s a very different equation.

    Textract Works Great with Forms

    What happens with most of these forms is that the OCR can’t really read the forms or actually make them coherent at all. Sometimes these templates will kind of effectively memorize in this box is this piece of data. Textract is going to work across legal forms and financial forms and tax forms and healthcare forms, and we will keep adding more and more of these.

    But also these forms will change every few years and when they do something that you thought was a Social Security number in this box turns out now not to be a date of birth. What we have built Textract to do is to recognize what certain data items or objects are so it’s able to tell this set of characters is a Social Security number, this set of characters is a date of birth, this set of characters is an address.

    Not only can we apply it to many more forms but also if those forms change Textract doesn’t miss a beat. That is a pretty significant change in your capability in being able to extract and digitally use data that are in documents.

  • Etsy CEO: Machine Learning is Opening Up a Whole New Opportunity

    Etsy CEO: Machine Learning is Opening Up a Whole New Opportunity

    Etsy CEO Josh Silverman says that “machine learning is opening up a whole new opportunity” for the company to organize 50 million items into a discovery platform that makes buying an enjoyable experience and also is profitable for sellers.

    Josh Silverman, CEO of Etsy, recently talked about their much-improved business and why it is working so well with Jim Cramer on CNBC:

    Our Mission is Keeping Commerce Human

    Our mission is keeping commerce human. It’s really about in a world where automation is changing the nature of work and we’re all buying more and more commoditized things from the same few fulfillment centers. Allowing someone to harness their creative energy and turn that creativity into a business and then connect with someone in the other part of the country or in another part of the world, that’s really special. We think there’s an ever-increasing need for that in this world.

    It’s about value. We’ve been really focused on delivering more value for our makers. Etsy really is a platform that brings buyers to sellers and that’s very valuable. We raised our commission from 3.5 to 5 percent commission which was I think is fair value for our sellers, particularly because we’re reinvesting 80 percent of that into the growth of the platform.

    Free shipping is pretty much table stakes today. Yet only about 20 percent of items have free shipping. About half of all the items on Etsy buyers say have shipping prices that are too high and yet we grew GMS at 20 percent last quarter.

    Machine Learning is Opening Up a Whole New Opportunity

    Machine learning is opening up a whole new opportunity for us to take 50 million items from two million makers and make sense of that for people. We have 37 million active buyers now and many of them come just for discovery, just to see what they can find, and that is exactly the right thing for someone out there. Our job is to create that love connection. Etsy over the past 14 years, with a large team effort, has I think done a great job.

    One thing I want to emphasize is the quality and the craftsmanship with so many of the products on Etsy. That’s something that has been such a delight for me. People like Kringle Workshops that make these incredible products. What we have been doing a better job and need to continue to do a better job of really surfacing the beautiful artisanally crafted products that are available at a really fair price. You’re not having to pay for warehousing, you’re not having to pay for all the other things that mass-produce things have to pay for, you’re buying directly from the person who made it. So it can be both beautiful, handcrafted, and well priced.

    There are 2 million sellers, 87 percent of them are women, over 90 percent are working from home or are businesses of one, who can create a global business from their garage or their living room. Etsy does provide a real sense of community for them and that’s really powerful.

    Amazon May Open New HQ in Queens Near Etsy

    We feel great about our employee value proposition and come what may. Here’s what we have going for us. We think we’ve got the best team, certainly in tech companies on the eastern seaboard. We think ours is the best and we continue to attract great talent. The reason is, first and foremost, our mission is really a meaningful important mission and that matters. Great people want to work in a place with a great mission.

    Second, our technology challenges are interesting. For example, search and using machine learning to make sense of 50 million items that don’t map to a catalog. Third, our culture is really special. We have been a company that’s authentically cared about diversity from the beginning. Over 50 percent of our executive staff are women, we have a balanced board, 50 percent male and female, and 32 percent of our engineers are female, which is twice the industry average. People who care about diversity and inclusion really want to come to work at Etsy. All of that is going for us and we’re happy to compete with whoever we need to.

    Earnings Call Comments by Etsy CEO:

    Active Buyers Grew 17 Percent

    Etsy’s growth accelerated again in the third quarter to nearly 21% on a constant-currency basis. Revenue growth exceeded 41%, fueled by the launch of our new pricing structure, and our adjusted EBITDA margins grew to nearly 23%, while we also increased our investments in the business.

    Active buyers grew 17% to 37 million worldwide. This is the fourth consecutive quarter that GMS has grown faster than active buyers, evidence that we are seeing increased buyer activity on the platform, which is a key proxy for improvement in frequency. We grew the number of active sellers by 8% and GMS per active seller is also increasing.

    Two principal levers contributed to our progress this past quarter. The first is our continued product investment, focused on improving the shopping experience on Etsy. By making it easier to find and buy the great products available for sale on Etsy, we’re doing a better job converting visits into purchases. The second lever was our new pricing structure, which enabled us to ramp up investments in marketing, shipping improvements and customer support.

    Successful Cloud Migration

    We achieved a significant milestone in our cloud migration this quarter, successfully migrating our marketplace, Etsy.com, and our mobile applications to the Google Cloud with minimal disruption to buyers and sellers. This increases our confidence that the migration will be complete by the end of 2019.

    Once fully migrated, we expect to dramatically increase the velocity of experiments and product development to iterate faster and leverage more complex search and machine learning models with the goal of rapidly innovating, improving search and ultimately driving GMS growth.

    In fact, we’re beginning to see some of those benefits today based on the systems we’ve already migrated. I’d like to thank our engineering team for their incredible work to get this – get us to this point.

     

  • How Adobe is Using AI to Transform the Customer Experience

    How Adobe is Using AI to Transform the Customer Experience

    Adobe has now integrated their artificial intelligence platform Adobe Sensei into Photoshop and most of their creative products. “Adobe Sensei is an AI and machine learning platform that deeply understands how our users work and delivers a lot of simple workflow that makes that magical moment happens in any of our applications,” noted Abhay Parasnis, CTO & EVP at Adobe. “What makes Sensei so unique is that Adobe is the only company in the industry that can marry art of content and creative expression and science of delight on a massive scale.”

    “The key areas we focus on are content intelligence, computational creativity, and the experience which is related to understanding events related to how content is delivered,” commented Scott Prevost, VP Engineering of Adobe Sensei and Search in an Adobe explanation of the product.

    “If I can go all the way from how I create content in the creative tool and then have the ability to personalize it at scale to Adobe Experience Cloud, then have the ability to measure it through analytics and feed the measurement back into the creative workflow, saying these designs work better, that actually is the holy grail in what customers tell us they want,” says Parasnis.

    Shantanu Narayen, Adobe CEO, recently commented on CNBC about how this is helping to improve the Adobe customer experience:

    On the creativity side, everybody fears the blank page, so if AI can start to infer what people want to do in terms of using either Photoshop or one of our creative products and when you can speak to the computer and it understands and infers what you want to do and makes our products and tools more accessible, that’s a huge win. Then you can attract a tremendous amount of customers.

    At the other end of the spectrum, when you have millions of customers hitting your website, the AI that we have on the Digital Experience Cloud being able to infer intelligence from the trillions of transactions and ensure that you get the right offer that was meant for you in real time, that’s something that humans cannot do.

    Those are two really good examples at different ends of the spectrum of how AI enables our customers to do more with our technology.

  • Machine Learning and It’s Impact on Search

    Machine Learning and It’s Impact on Search

    The terms machine learning (ML) and artificial intelligence (AI) have been cropping up more often when it comes to organic and paid search. Now a recent report by Acquisio has confirmed just how effective machine learning is for search results.

    According to Acquisio, paid search accounts that have been optimised for machine learning have 71% higher conversion rates and have lower cost-per-click (CPC). But these were not the only benefits that accounts using machine learning enjoyed. The web marketing company also revealed these accounts were also able to reach their target spending levels and had lower churn rates.

    The data implies that small marketing teams and CMOs now stand on an even playing field with more established companies now that ML is more affordable, effective and accessible to everyone.

    This doesn’t mean that marketers should ignore organic search and original, value-laden content. Paid search might be the easiest way to rank high in search engines, particularly since AI will be doing the bulk of the work, developing campaigns that have greater odds of being seen by the right searchers at the proper time. However, organic search is more authentic and will last longer than paid searches.

    The goal now is to understand how ML impacts the search system and how to take advantage of the technology’s evolution that made paid and organic searches more effective.

    Paid vs Organic Search: Which Wins in the End?

    There’s been an ongoing debate as to which is better – paid or organic searches. Interestingly, both have come out on top, but at different times and conditions. The results have depended on the type of research done and other outside factors. For instance, a study conducted in 2011 showed that organic search was more effective. However, paid search has outpaced its counterpart from 2013 onwards. But this appears to be due to the changes Google has made to its algorithm.

    So which is better? Andy Taylor, the Associate Director of Research at Merkle, believes that flexibility is the best option. Instead of just sticking to one approach, companies should determine what search strategy is ideal for their business at the moment and the technology that’s currently available. After all, the ideal marketing strategy for your company now will probably change in a few months as customers change their expectations and technologies expand.

    Machine Learning is Changing More Than Search

    The rise of machine learning has also resulted in a shift to data-driven models instead of the conventional attribution models. This multi-touch attribution model (MTA) relies on an analytics scale that’s more descriptive and takes into account various touchpoint outputs, like ad interactions, ad creative, or exposure order. It also allows marketers to have a better understanding of how factors, like a distinct set of keywords and ad words, can affect a conversion.

    But it’s not just search capacities that machine learning has an impact on. The technology is also being used to refine and make algorithm changes. It has been theorized that Google’s RankBrain utilizes machine learning to assess if the company has to revise its own rankings based on what the consumer searches for and whether the user was satisfied with the result.

    Machine Learning Will Push for More Sophisticated Content

    Because machine learning technology is developing more advanced SEM capacities and sophisticated algorithms, search engines are pushing marketers and content producers to deliver more refined content. This would eventually lead to search engines becoming more discerning to the quality of online content a company is putting out. This means producing high-quality content that particularly targets what the consumer is looking for becomes more vital than ever before.

    Machine learning and AI are impacting every aspect of marketing. Companies should start understanding them and how to utilize ML-optimized tools effectively in their marketing campaigns.  

  • Facebook Combats Fake News With Related Articles and Updated Machine Learning Technology

    Facebook Combats Fake News With Related Articles and Updated Machine Learning Technology

    Facebook is making steps to ensure that everyone gets their facts right before forming their own opinion on a topic trending on the popular social media platform. Recently, the company tweaked its Related Articles feature and introduced an updated machine learning technology to help spot potential misleading or fake news.

    Facebook’s Related Articles feature, which is basically a list of recommended articles appearing after a post, was introduced in 2013 as a way to expose readers to other viewpoints from different sources on the same topic. With the additional information and perspective, it was hoped that readers of trending social media posts will become better informed on their topics of interest, allowing them to participate intelligently in online discussions. In addition, it is also a clever way of minimizing the spread of fake news as readers will be able to do some fact-checking themselves before sharing a story.

    After the company was accused of enabling the spread of fake political news on its platform during the U.S. elections, Facebook tweaked the Related Articles feature last April. The updated feature, which was tested to a small group of U.S. users, now shows the list of additional articles to readers even before they read an article in the news feed as they appear as links below a news story in their feeds.

    Facebook announced that the updated feature is now being rolled out broadly in the United States as of August 3, 2017. Forbes reported that the updated Related Articles feature is also available for users in some European countries such as the Netherlands, France, and Germany.

    Recently, fake news has become a tricky issue for Facebook in Europe as well. Before the May 2017 presidential election in the country, French voters were reported to have been bombarded with a barrage of fake news reports. With its election coming this September, Germany is mulling over a plan to fine social network platforms should they fail to halt the spread of hateful posts.

    But Facebook has technology on its side. The social media giant likewise announced that it can now screen potential fake news better with its updated machine learning technology.

    [Featured Image by Pixabay]

  • Apple Shares Source Code For Machine Learning Framework at WWDC 2017

    Apple Shares Source Code For Machine Learning Framework at WWDC 2017

    Apple’s recent WWDC (Worldwide Developers Conference) saw the unheralded release of Core ML, which will reportedly make it easier for developers to come up with machine learning tools across the Apple ecosystem.

    The way this works is that developers need to convert their creations into an API that is compatible with the Core ML. They then have to load their programs into the Apple Xcode development before it can be installed on the iOS.

    Developers can use any of the following frameworks: Keras, XGBoost, LibSVM, Caffe, and scikit-learn. To make it even easier for them to load their models, Apple is allowing them to come up with their own converter.

    According to Apple, the Core ML is “a new foundational machine learning framework used across Apple products, including Siri, Camera, and QuickType.”

    The company explained that this new machine learning tool would be “the foundation for domain-specific frameworks and functionality.”

    One of the primary advantages of the Core ML is that it speeds up the artificial intelligence on the Apple Watch, iPhone, iPad, and perhaps the soon-to-be-released Siri speaker. If it works the way that is billed, any AI task on the iPhone, for instance, would be six times quicker than Android.

    The machine learning tools supported by Apple Core ML include linear models, neural networks, and tree ensembles. The company also promised that private data by users won’t be compromised by this new endeavor. This means that developers can’t just tinker with any phone to steal private information.

    Core ML also supports:

    • Foundation for Natural Language Processing
    • Vision for Image Analysis
    • Gameplay Kit

    “Core ML itself builds on top of low-level primitives like Accelerate and BNNS, as well as Metal Performance Shaders,” the company added.

    But Apple is reportedly not content with just releasing the Core ML. According to rumors, the company is looking to fulfill its promise of helping to build a very fast mobile platform. In fact, if the rumors are true, the company is also building a much better chip that can handle AI tasks without compromising performance.

    Though Core ML seems promising, Apple is certainly not blazing the trail when it comes to machine learning. In fact, Facebook and Google have already unveiled their own machine learning frameworks to optimize the mobile user’s experience.

    The new machine learning framework is still part of Apple’s Core Brand, which already includes Core Audio, Core Location, and Core Image as announced earlier.

  • Mark Zuckerberg on AI, Poking, and Whether or Not the Machines Win

    Mark Zuckerberg on AI, Poking, and Whether or Not the Machines Win

    Facebook CEO Mark Zuckerberg recently held another Q&A, in which he talked artificial intelligence, virtual reality, and the future of the company. He also had nice interchanges with Stephen Hawking and Arnold Schwarzenegger, and answered a question about Poking.

    Here are some of his most interesting responses.

    On the topic of Facebook’s “real name policy” and its effect on the transgender community:

    Real names are an important part of how our community works for a couple of reasons.

    First, it helps keep people safe. We know that people are much less likely to try to act abusively towards other members of our community when they’re using their real names. There are plenty of cases — for example, a woman leaving an abusive relationship and trying to avoid her violent ex-husband — where preventing the ex-husband from creating profiles with fake names and harassing her is important. As long as he’s using his real name, she can easily block him.

    Second, real names help make the service easier to use. People use Facebook to look up friends and people they meet all the time. This is easy because you can just type their name into search and find them. This becomes much harder if people don’t use their real names.

    That said, there is some confusion about what our policy actually is. Real name does not mean your legal name. Your real name is whatever you go by and what your friends call you. If your friends all call you by a nickname and you want to use that name on Facebook, you should be able to do that. In this way, we should be able to support everyone using their own real names, including everyone in the transgender community. We are working on better and more ways for people to show us what their real name is so we can both keep this policy which protects so many people in our community while also serving the transgender community.

    On the future of technology:

    In 10 years, I hope we’ve improved a lot of how the world connects. We’re doing a few big things:

    First, we’re working on spreading internet access around the world through Internet.org. This is the most basic tool people need to get the benefits of the internet — jobs, education, communication, etc. Today, almost 2/3 of the world has no internet access. In the next 10 years, Internet.org has the potential to help connect hundreds of millions or billions of people who do not have access to the internet today.

    As a side point, research has found that for every 10 people who gain access to the internet, about 1 person is raised out of poverty. So if we can connect the 4 billion people in the world who are unconnected, we can potentially raise 400 million people out of poverty. That’s perhaps one of the greatest things we can do in the world.

    Second, we’re working on AI because we think more intelligent services will be much more useful for you to use. For example, if we had computers that could understand the meaning of the posts in News Feed and show you more things you’re interested in, that would be pretty amazing. Similarly, if we could build computers that could understand what’s in an image and could tell a blind person who otherwise couldn’t see that image, that would be pretty amazing as well. This is all within our reach and I hope we can deliver it in the next 10 years.

    Third, we’re working on VR because I think it’s the next major computing and communication platform after phones. In the future we’ll probably still carry phones in our pockets, but I think we’ll also have glasses on our faces that can help us out throughout the day and give us the ability to share our experiences with those we love in completely immersive and new ways that aren’t possible today.

    Those are just three of the things we’re working on for the next 10 years. I’m pretty excited about the future

    On Facebook’s AI initiatives:

    Most of our AI research is focused on understanding the meaning of what people share.

    For example, if you take a photo that has a friend in it, then we should make sure that friend sees it. If you take a photo of a dog or write a post about politics, we should understand that so we can show that post and help you connect to people who like dogs and politics.

    In order to do this really well, our goal is to build AI systems that are better than humans at our primary senses: vision, listening, etc.

    For vision, we’re building systems that can recognize everything that’s in an image or a video. This includes people, objects, scenes, etc. These systems need to understand the context of the images and videos as well as whatever is in them.

    For listening and language, we’re focusing on translating speech to text, text between any languages, and also being able to answer any natural language question you ask.

    This is a pretty basic overview. There’s a lot more we’re doing and I’m looking forward to sharing more soon.

    From Stephen Hawking:

    I would like to know a unified theory of gravity and the other forces. Which of the big questions in science would you like to know the answer to and why?

    That’s a pretty good one!

    I’m most interested in questions about people. What will enable us to live forever? How do we cure all diseases? How does the brain work? How does learning work and how we can empower humans to learn a million times more?

    I’m also curious about whether there is a fundamental mathematical law underlying human social relationships that governs the balance of who and what we all care about. I bet there is.

    From Arnold Schwarzenegger:

    Mark, I always tell people that nobody is too busy to exercise, especially if Popes and Presidents find time. You’ve got to be one of the busiest guys on the planet, and younger generations can probably relate to you more than they can the Pope – so tell me how you find time to train and what is your regimen like? And by the way – will the machines win?

    Staying in shape is very important. Doing anything well requires energy, and you just have a lot more energy when you’re fit.

    I make sure I work out at least three times a week — usually first thing when I wake up. I also try to take my dog running whenever I can, which has the added bonus of being hilarious because that basically like seeing a mop run.

    And no, the machines don’t win

    On his definition of happiness:

    To me, happiness is doing something meaningful that helps people and that I believe in with people I love.

    I think lots of people confuse happiness with fun. I don’t believe it is possible to have fun every day. But I do believe it is possible to do something meaningful that helps people people every day.

    As I’ve grown up, I’ve gained more appreciation for my close relationships — my wife, my partners at work, my close friends. Nobody builds something by themselves. Long term relationships are very important.

    On why he has a $1 set salary at Facebook:

    I’ve made enough money. At this point, I’m just focused on making sure I do the most possible good with what I have. The main way I can help is through Facebook — giving people the power to share and connecting the world. I’m also focusing on my education and health philanthropy work outside of Facebook as well. Too many people die unnecessarily and don’t get the opportunities they deserve. There are lots of things in the world that need to get fixed and I’m just lucky to have the chance to work on fixing some of them.

    On the future of Facebook:

    There are a few important trends in human communication that we hope to improve.

    First, people are gaining the power to share in richer and richer ways. We used to just share in text, and now we post mainly with photos. In the future video will be even more important than photos. After that, immersive experiences like VR will become the norm. And after that, we’ll have the power to share our full sensory and emotional experience with people whenever we’d like.

    Second, people are gaining the power to communicate more frequently. We used to have to be with someone in person. Then we had these bulky computers at our desks or that we could carry around. Now we have these incredible devices in our pockets all the time, but we only use them periodically throughout the day. In the future, we’ll have AR and other devices that we can wear almost all the time to improve our experience and communication.

    One day, I believe we’ll be able to send full rich thoughts to each other directly using technology. You’ll just be able to think of something and your friends will immediately be able to experience it too if you’d like. This would be the ultimate communication technology.

    Our lives improve as our communication tools get better in many ways. We can build richer relationships with the people we love and care about. We know about what’s going on in the world and can make better decisions in our jobs and lives. We are also more informed and can make better decisions collectively as a society. This increase in the power people have to share is one of the major forces driving the world today.

    And finally …

    Why did you come up with Poking?

    It seemed like a good idea at the time.

    It always does, Mark. It always does.

  • Amazon Gives Its Machine Learning Capabilities To Developers

    Amazon Gives Its Machine Learning Capabilities To Developers

    Amazon is giving its machine learning capabilities to developers as an Amazon Web Services offering. According to the company, it makes it easier for developers to use historical data to build and deploy predictive models, which can be used for things like detecting problematic transactions, preventing customer churn, and improving customer support.

    The offering is based on the same machine learning technology Amazon’s own developers use, which the company says includes generating over 50 billion predictions a week.

    Amazon Machine Learning hooks developers up with APIs and wizards that guide them through creating and tuning machine learning models that can be “easily deployed and scale to support billions of predictions.”

    It’s all integrated with Amazon Simple Storage Service (S3), Amazon Redshift, and Amazon Relational Database Service (RDS), so it will work with data that’s already stored in the AWS cloud.

    “Until now, very few developers have been able to build applications with machine learning capabilities because doing so required expertise in statistics, data analysis, and machine learning,” Amazon explains. “In addition, the traditional process for applying machine learning involves many manual, repetitive, and error-prone tasks such as computing summary statistics, performing data analysis, using machine learning algorithms to train a model based on data, evaluating and fine tuning the model, and then generating predictions using the model. Amazon Machine Learning makes machine learning broadly accessible to all software developers by abstracting away this complexity and automating these steps. With Amazon Machine Learning, developers can use the AWS Management Console or APIs to quickly create as many models as they need, and generate predictions from them with high throughput without worrying about provisioning hardware, distributing and scaling the computational load, managing dependencies, or monitoring and troubleshooting the infrastructure. There is no setup cost, and developers pay as they go so they can start small and scale as an application grows.”

    “Amazon has a long legacy in machine learning. It powers the product recommendations customers receive on Amazon.com, it is what makes Amazon Echo able to respond to your voice, and it is what allows us to unload an entire truck full of products and make them available for purchase in as little as 30 minutes,” adds Jeff Bilger, Senior Manager of Amazon Machine Learning. “Early on, we recognized that the potential of machine learning could only be realized if we made it accessible to every developer across Amazon. Amazon Machine Learning is the result of everything we’ve learned in the process of enabling thousands of Amazon developers to quickly build models, experiment, and then scale to power planet-scale predictive applications.”

    Amazon Machine Learning lets developers visualize statistical properties of datasets that will be used to train the model to find data patterns, which the company says saves time by helping devs understand and identify data distributions and missing or invalid values before actually training the model. The training data is automatically transformed, and the machine learning algorithms are optimized so the developers “don’t need a deep understanding” of such algorithms or tuning parameters to come up with the best possible solution.

    The offering also includes built-in quality alerts to help developers build and refine models.

    According to Amazon, in 20 minutes, one developer was able to use the technology to solve a problem that had taken two developers 45 days to solve. None of them had any experience in machine learning.

    Comcast is already using Amazon Machine Learning for data science analytics.

    “We particularly liked the ability to visually explore the tradeoff between parameter settings and classification performance during the evaluation,” said Jan Neumann, Manager of a Data Science Research team at Comcast. “With Amazon Machine Learning it was quite simple to prepare and clean the input data and train a model on large data sets in short order.”

    Now, if they can just come up with an algorithm for customer service.

    On the Amazon Machine Learning site, the company lists the following as popular use cases: fraud detection, document classification, content personalization, customer churn prediction, propensity modeling for marketing campaigns, and automated solution recommendation for customer support.

    Image via YouTube

  • Twitter Acquires Machine Learning Company Madbits

    Twitter Acquires Machine Learning Company Madbits

    Twitter has made another acquisition, picking up Madbits, which specializes in deep learning/machine learning, and has been working on technologies that can help extract information from raw media, such as images.

    This is the kind of thing that a company like Twitter could obviously put to good use with the amount of such raw media that saturates the service every day, or every second for that matter.

    Madbits talks about joining Twitter on its site (via GigaOm, which first reported on the acquisition).

    Over this past year, we’ve built visual intelligence technology that automatically understands, organizes and extracts relevant information from raw media. Understanding the content of an image, whether or not there are tags associated with that image, is a complex challenge. We developed our technology based on deep learning, an approach to statistical machine learning that involves stacking simple projections to form powerful hierarchical models of a signal.

    We prototyped and tested about ten different applications, and as we’ve prepared to launch publicly, we’ve decided to bring the technology to Twitter, a company that shares our ambitions and vision and will help us scale this technology.

    We are excited to join the folks at Twitter to merge our efforts and see this technology grow to its full potential.

    Twitter also mentioned on its earnings call on Tuesday that it has completed its previously announced acquisition of Tap Commerce. The company also announced a couple weeks ago that it has acquired CardSpring.

    Terms of the deal were not disclosed.

    Image via Twitter

  • Bing Improves Image Search With Deep Learning

    Bing Improves Image Search With Deep Learning

    Image search is a cornerstone of any search engine. That’s why both Google and Bing are doing everything they can to improve image search to bring up the most relevant images for any search imaginable. While some may argue that recent changes made to Google image search make it worse, Bing is moving ahead with a new strategy that involves deep learning.

    So, what is deep learning? In short, it’s a type of machine learning that uses artificial neural networks to learn about and understand multiple concepts, including the abstract. In the past, computer systems had to be manually “trained” to recognize patterns or specific images. With machine learning, these systems can now learn to recognize these patterns on their own.

    When it comes to image search quality, Bing found that integrating deep learning into its systems greatly increased the quality. With deep learning enabled, a search for cats returns all cats except for two dogs that happen to look like cats. Using traditional search features, the search returns only two cats with the rest of the results featuring dogs, a baby and a disembodied head.

    In short, Bing hopes to use deep learning to provide better search results by connecting like images via a giant graph. Here’s the full explanation:

    Two images can be connected if the distance between the respective features learned through deep learning is small enough. Extending this concept to all the images on the web, trillions of connected images form a gigantic graph where each image is connected via semantic links to other images. As illustrated in the graph below, by using deep learning features, the image of a motorcycle is connected with other images with motorcycles of different colors and shapes. By using traditional features such as colors and edges, the same image of a motorcycle is connected to images of different entities such as bicycles, or even waterfalls and landscapes. In contrast, deep learning keeps the semantics in the image neighborhood even though the visual patterns are not very similar.

    The above might be a little confusing to understand so here’s the above concept in visual form:

    Bing Improves Image Search With Deep Learning

    As you can see in the first image, all the connected images are of motorcycles. They may not be similar motorcycles, but the system recognizes that a person is searching for a motorcycle. In the bottom image, the search results are a little more chaotic as it returns some motorcycles, but it also returns images of waterfalls and bicycles simply because the images are similar in color, among other indicators.

    With deep learning enabled, Bing should be able to return more relevant images in search than before. It probably won’t fix its suggested image search problem though.

    [Image: Bing]

  • Bing Uses Machine Learning To Improve Product Search

    Bing Uses Machine Learning To Improve Product Search

    Bing announced today that it is launching a new product search experience, which utilizes its index of tens of millions of products, along with machine learning, to show products on the main results page, rather than making users go to a dedicated shopping page.

    As you can see, it looks vaguely like Google’s Knowledge Carousel and the style the search engine recently added to local results.

    Bing Product Search

    “For example, if you are in the market for a new camera. Simply enter ‘dslr camera’ and you will see a category snapshot with the top dslr cameras,” Bing explains in a blog post. “You might have heard about a specific Canon model, you click on the <Canon EOS 6D> and you will see our new Carousel so you can quickly browse other related products that might be of interest and then narrow down the selection that best meets your needs. We’re also bringing the power of our industry leading Snapshot technology to expose more of the details you need for a specific product, like specs and reviews, to make a decision without having to click to another page.”

    “If you want to purchase the product, we present you with two easy options,” Bing says. “First, product pricing and availability directly from the merchant themselves will appear right inside search results in what we call Rich Captions. So if Target, for example, has the product you’re looking for, and that link appears in search results, Bing will show you their price and availability information. Another way Bing helps is through new product adswith photos and pricing that provide you the ability to quickly see offers from merchants across the web.”

    The new product search experience will be rolling out over the course of the coming months, so you may still see the old version for a while.

    Image: Bing

  • Microsoft Talks Machine Learning, Biological Science

    Microsoft Talks Machine Learning, Biological Science

    David Heckerman from Microsoft research recently discussed some examples of how machine learning is affecting genomics and changing the pace of scientific breakthroughs.

    “Until recently, the wet lab has been a crucial component of every biologist,” Microsoft Research says. “Today’s advances in the production of massive amounts of data and the creation of machine-learning algorithms for processing that data are changing the face of biological science—making it possible to do real science without a wet lab.”

    Microsoft just released the video Heckerman’s talk. It’s just over a half hour long:

  • Diffbot Makes The Web Machine-Readable

    Diffbot Makes The Web Machine-Readable

    Microsoft’s Bing service has a cool promotion vehicle known as Bing Booster. It helps tech startups connect and collaborate with other startups and people in their field. One of the events Bing helps sponsor is LAUNCH, an event that give startups a platform to launch from.

    One of the startups that Bing brought to LAUNCH this year is really interesting. The company is called diffbot, and they have a mission. They want to make the entire Web machine-readable. What does that mean? According to the official Web site, diffbot is a “visual learning robot that enables developers to easily use Web content in their apps.”

    So diffbot lets a robot read the Web regardless of layout, design or language. That’s pretty cool, but what kind of applications would it have for the Web. The diffbot team lays out its potential uses on the BingBooster Web site:

    Using Diffbot’s existing Article API (which automatically parses blog posts or news articles into machine- app-friendly XML or JSON) to migrate users’ blogs — regardless of existing platform — to a new blog-platform provider.

    Providing Diffbot’s forthcoming Product API (which can parse product pages of any type across the web) with additional training data or edge cases from existing product-search and price-comparison applications.
    Leveraging Diffbot’s language agnosticism to help power a multi-language tablet newsreader focused on global content and stories.

    Powering a forthcoming event and activity search engine using Diffbot’s impending events page functionality.

    Using our image-identification and extraction capabilities to power a personal photo-book printing service, allowing the inclusion of photos from any gallery software or photo-hosting services.

    The LAUNCH event was a great chance for the diffbot team to meet various people that were interested in using their API as well as helping them find new uses for their technology.

    The BingBooster Web site says this is just the first in a series of startups that showed off their product at LAUNCH. There will be more development technologies being shown during the coming days.

    Here’s a presentation of diffbot at the DEMO Enterprise Disruption 2012 from a few months ago:

  • Todd Tweedy, CEO Of Audience Machine, Missing After Suicidal Facebook Post [updated]

    Todd Tweedy, CEO Of Audience Machine, Missing After Suicidal Facebook Post [updated]

    Update: Tweedy has reportedly been found alive.

    Todd Tweedy, CEO of Audience Machine and author of an upcoming Wiley book Facebook Marketing Secrets is missing. The last known communication from him is a status update on Facebook on Tuesday, which said:

    One illness I’ve never been able to defeat is my own depression. I have to say goodbye now. I wish each of you a wonderful New Year!

    His bio at Audience Machine says:

    As CEO of Audience Machine, Todd Tweedy is responsible delivering technology-enabled community marketing and search engine marketing services to achieve cost-effective online customer acquisition and revenue goals.

    Mr. Tweedy’s client experience includes Sylvan Learning, Rolex, AOL, LikeMe, MTV, Microsoft, Metagenics, Netscape, MTV Networks, Quest Communications, Discovery Channel, FedEx, US West, US Airways, Volkswagen, Audi and Scholastic among many others.

    Mr. Tweedy has authored numerous articles on ecommerce and email as well as the word of mouth industry study Perceptions, Practices & Ethics in Word-of-Mouth Marketing that has been downloaded over 150,000 times since being published in May of 2006. Mr. Tweedy also authored The Neighboring Marketing Modelpublished by Internet.com in 2002 on how to leverage public instant messaging networks to support acquisition marketing.

    Todd is a big-picture technology-driven marketing geek with documented ability to drive strong acquisition growth and online sales. Vast e-commerce experience from airline reservation systems to book sales and from affiliate marketing to viral and social networking over 16 years with B2C & B2B leadership for startups, ad agencies as well as Fortune 500 firms.

    He lives in St. Paul Minnesota with his wife and three children, according to that bio. A CBS Minnesota report (via Marketing Land) says the Bureau of Criminal Apprehension is involved in the search Tweedy. It says:

    The Bureau of Criminal Apprehension described Tweedy as 6 feet, 3 inches tall with brown hair and brown eyes. The BCA said Tweedy may be driving a red Volkswagen Passat with the license place 545 BLM.

    Anyone with information about his whereabouts is asked to contact St. Paul Police at 651-266-5612.

    Many have chimed into comment on his Faceboo posts, showing their support and offering help – including people who claim not to know him personally.

    On Twitter, the hashtag #FindTodd is being used to spread awareness and to help locate him.

    @StephenCaggiano
    Stephen J. Caggiano    ALERT! Please help find Todd Tweedy. Missing since 1/3 Drives Red Passat, MN 545-BLM http://t.co/UN1UCdmt #findtodd to help cc/ @ty_sullivan 5 minutes ago via TweetDeck · powered by @socialditto

  • Google Offers $5.7 Million In Research Funding

    Google Offers $5.7 Million In Research Funding

    Google said today it is increasing its funding for 12 university projects to further advance research in areas the company is interested in developing.

    Google Focused Research Awards, totaling $5.7 million, include four categories: machine learning, the use of mobile phones as data collection devices for public health and environment monitoring, energy efficiency in computing, and privacy.
    Aflred-Spector.jpg
    "These are all areas in which Google is already deeply invested, and yet there is a long way to go. We’re excited to see what these projects contribute to the body of research in these important areas," wrote Alfred Spector, Vice President of Research and Special Initiatives, Google.

    "These unrestricted grants are for two to three years, and the recipients will have the advantage of access to Google tools, technologies and expertise."
     

    Related Articles:

    > Google Profiles Go to the SERPs

    > Google Launches Social Search Experiment

    > Can Search Engine Optimization Survive Google?