Diese Präsentation wurde erfolgreich gemeldet.
Wir verwenden Ihre LinkedIn Profilangaben und Informationen zu Ihren Aktivitäten, um Anzeigen zu personalisieren und Ihnen relevantere Inhalte anzuzeigen. Sie können Ihre Anzeigeneinstellungen jederzeit ändern.

Summary artificial intelligence in practice- part-3

243 Aufrufe

Veröffentlicht am

This is Part-2 . Read this after reading Part 1 & 2

Veröffentlicht in: Technologie
  • Get Paid For Your Opinions! Earn $5-$10 cash on your first survey. ★★★ http://ishbv.com/goldops777/pdf
    Sind Sie sicher, dass Sie …  Ja  Nein
    Ihre Nachricht erscheint hier

Summary artificial intelligence in practice- part-3

  1. 1. Some Impressionistic Take away from the Book of Bernard Marr & Matt Ward Artificial Intelligence in Practice ( Part – 3) ( How 50 Successful companies used AI & Machine Learning to Solve problems) Ramki ramaddster@gmail.com
  2. 2. The Summary of this book is made in 4 parts due to large coverage of the book . This is Part – 3 ( Read this after Part-1 & 2)
  3. 3. Part – 3 Media, Entertainment & Telecom Companies
  4. 4. Using AI to make Magical Memories
  5. 5. Walt Disney  Disney is known for fiercely protecting the magic behind the curtain that makes the “Most Magical Place on Earth” hum.  Even though we may not know all their secrets, we do know Disney masterminds are leveraging Big Data in innovative ways to improve the experience of those who walk through the gates of their parks.  There are also some intriguing developments with using data that will excite Disney movie fans.  Let’s take a look at how big data is changing the entertainment behemoth of Disney. The Magic of MagicBands  In 2013, after years of development and testing, Disney World launched its MyMagicPlus program.  Now, every guest to Disney World gets a MagicBand, a wristband that is equipped with RFID technology and a long-range radio.  These bands communicate with thousands of sensors and stream real-time data to hundreds of systems that make the entertainment venue a giant computer.  All this data is designed to help Disney cast members anticipate all your desires so they can give you an incredible experience.  The bands act as hotel keys, credit cards, tickets, FastPasses and more. With a simple swipe of the band across sensors located throughout the park, the giant system knows where you are, what you’re doing and what you need.
  6. 6. Walt Disney  The goal of the tech team who developed the MagicBands was to “root out all the friction within the Disney World experience.”  Even before you leave town you can set reservations for certain attractions (where you won’t have to wait in line—hallelujah!) And added bonus for Disney: Your choices get added to its data vault.  Once you arrive on site, one of the biggest challenges of any amusement park is how to minimize the wait times for rides and attractions.  When people are waiting in line, they aren’t spending money on food or shopping.  As each guest swipes their band at a ride, vital intel is being shipped real- time to the operations team.  This allows decisions to be made about adding staff or incentivizing guests to head to another ride or attraction.  This re-rerouting of guests makes more efficient use of the park and even allows for exemplary customer service to be delivered.  As data fuels a better experience, the trepidation and creepiness of machines knowing so much about you, melts away as families experience a trip of a lifetime.
  7. 7. Walt Disney  Imagine all the possibilities. A family pre-orders dinner from their hotel room, and when they arrive at the restaurant not only are they greeted enthusiastically—and by name—their food is instantly delivered. All of this transpires because the system triangulates the family’s location and alerts the wait staff of their arrival.  The possibilities for improving the experience of Disney guests are endless.  Your favorite Disney character could find you and greet your child by name.  Candid photos of your family enjoying the park can be taken throughout the day and sent to your hotel room each night.  What if you wait too long in a line?  The system knows and could deliver you a free voucher for your trouble.  That’s a sure-fire way to turn a frown upside down; figuring out how to turn a negative experience into a positive one is the holy grail of customer service.
  8. 8. Magic Band
  9. 9. Results, Key Challenges, Learning Points & Takeaways Visitors to the Magic Kingdom can cram a lot more into their day & go home with better memories ( more souvenirs & merchandise) if the “ friction” can be removed from their trip. Less waiting in line leads to happier customers who are more likely to return. Disney uses its theme parks to bring it characters & movies into its visitors’ real lives. It hopes this will make them bond more closely with its brands and franchises, and continue to buy its movies and tie-in merchandise. Advanced, intelligent analytics can vastly simplify managing the movement of huge numbers of people. Risk averse through technology . Overcoming this was a challenge for the team that built the MyMagic+Initiative.
  10. 10. Using AI to tackle online Bullying
  11. 11. Instagram  Instagram, the social networking app for sharing photos and videos, launched in 2010.  Today, it boasts 800 million monthly active users and is owned by Facebook- 1 billion active users as of June 2018-  There are 95 million photos uploaded to Instagram every day.  People interact with each of those posts by showing their love with a heart, commenting and using hashtags.  What all of this activity does is create an enormous amount of data.  Once analyzed, by humans as well as increasingly through artificial intelligence algorithms, it can provide incredible business intel and insights into human behavior causing Instagram CEO Kevin Systrom to say “We’re also going to be a big data company.”  Here are some ways Instagram uses big data and artificial intelligence today. Explore Page and Search Function  Via the use of tags and trending information, Instagram users are able to find photos for a particular activity, topic or event or discover experiences, restaurants and places around the world that are trending.  Enabled by tagging, the search tools help Instagram users discover things of interest among the millions of uploaded images.
  12. 12. Instagram Target Advertising  In order to make the data that Instagram collects valuable, it must extract customer insights from it.  By assessing the search preferences and engagement insights from its users, Instagram can sell advertising to companies who want to reach that particular customer profile and who might be most interested in receiving a particular marketing message.  Since Facebook with 1.8 billion users owns Instagram they have a powerful network of analytics information to help target advertising based on what people like, who they follow and interact with and what they save. Enhance the User Experience  In order to ensure users find value in the platform, it’s important for Instagram to show them what they will like.  As the amount of content grows, finding content that each user will find relevant becomes exponentially more challenging.  When Instagram changed its feed from reverse-chronological order to showing posts that they believe users would like and share, machine-learning algorithms were put on the job to help sort the information and to better learn over time what is most valued and relevant for each user to create a personalized feed.
  13. 13. Instagram Filter Spam  Instagram uses artificial intelligence to fight spam. The spam filter is able to remove fake messages from accounts written in nine languages including English, Chinese, Russian, Arabic and more. Once messages are detected they are automatically removed. Instagram uses Facebook’s artificial intelligence text analytics algorithm DeepText that is able to understand the context of a message nearly as good as humans. Fight Cyberbullying and Delete Offensive Comments  In a survey conducted by Ditch the Label, 42% of more than 10,000 UK youth between ages 12 and 25 reported Instagram was the platform where they were most bullied.  With this unfortunate distinction of having the biggest cyberbullying problem of any social media site, they became the first to use machine learning to automatically remove offensive posts, whereas Facebook and Twitter rely on users to report abusive language.  Based on the success of using DeepText to identify spam and remove it, Instagram officials began to see it as a solution to identify and eliminate comments that violate Instagram’s Community Guidelines.  Humans reviewed and tagged actual Instagram posts to help DeepText learn what would be considered offensive content in certain contexts and what wouldn’t be. If the algorithm finds something offensive, it is immediately removed.
  14. 14. Instagram Study the Human Condition  In one study, 100 million Instagram photos were used to learn global clothing patterns.  What would have been an impossible amount of data to review is increasingly possible thanks to machine learning.  This work showed the potential for machine learning to help extract insights when studying humans and social, economic and cultural factors around the world.  This study exhibited the power of big data and technologies such as computer vision, automated analysis algorithms and machine learning to comb through enormous data sets that are created by social media sites to gain understanding about the world around us.  From enhancing its platform for users and advertisers to finding and removing fake or offensive content, Instagram uses the insights it extracts from all the data it collects to improve while others find great potential in the enormous data it collects to uncover insights about human behavior, cultures and more.
  15. 15. Results, Key Challenges, Learning Points & Takeaways Instagram anti-bullying initiative is very new & the company has not spoken about the results it has seen yet. Bullying is a challenge that is always existed in society, but the internet & social media make matters worse as victims can be targeted publicly & anonymously. Without AI it was not possible to screen every upload to Instagram in real time. This meant that proactive blocking as enabled by Deeptext would not be possible. Text analytics & natural language processing are advanced enough now to reliably make the right call.
  16. 16. Using AI to solve the Skills Crisis
  17. 17. Linkedin  What Facebook has done in keeping touch with family & friends, Linkedin replicates for working lives.  Revenue – FB – makes money by selling our data to businesses which helps in advertisement whereas Linkedin revenue comes from employers looking to entice people to join their organization.  What Problem AI helping to solve  Matching applicants to job roles is a challenging & expensive task for the businesses.  Average cost of hiring in USA to a company is US $ 4000.  Human beings are not particularly great at selecting the right person for the job- 2 out 5 is the best number.  Unsuccessful hire costs a company on an average UK 132,000.  Part of the inefficiency in recruiting is gathering information during the process.  One million more nurses will need to recruited by 2024- aging problem.  Technology facing skill crisis.  IBM data – 2.7 million unfilled vacancies in Data science by 2020
  18. 18. Linkedin- How AI is used ?  Linkedin gathers data on millions of professionals & then uses AI search tools to match applicants with job , and vice versa.  Builds network seamlessly – uses AI to identify these connections.  AI suggests courses to benefit the users from its library.  Can mark “ Open candidates” meaning they are open to new assignments.  Tracks the jobs the user browses  Matches with the employer requirements through algorithms.  System uses Machine learning – continuously refining algorithms based on feedback from the past.  More predictive in terms of candidates to fill the job role – unlikely noticed by a human being  Machine learning will build patterns in the links between candidates which increases the confidence in its predictions.  Data driven insights are used to determine what features and functionality users will get from the platform for the future.
  19. 19. Technology  Linkedin uses data that users give in the social media on professional lives.  It builds a picture of which vacancies might tempt them to respond & categorize them.  It also builds up profiles as they use the service by monitoring what companies & vacancies they browse and who joins their personal network.  From employers data is collected on its users from the recruiter platform.  This data is used to profile recruiters & build models that predict what they are looking for in recruits.
  20. 20. Results, Key Challenges, Learning Points & Takeaways Improvements to the AI algorithms used in its tools & search engines – increased response rate to its users InMail by 45%. LinkedIn fuel is data its users feed it and what it can learn from their behavior as they use its services. Machine learning algorithms can be used to accurately match candidates with job vacancies, but they need the data to be able to do it. AI job matching can encourage employers to consider candidates with experience and skillsets that differ from the preconceptions about what a job requires.
  21. 21. Using AI to Give us a better TV experience
  22. 22. Netflix  Netflix – evolved from DVD-by mail rental company to become a subscription based streaming video-on-demand service- 130 million subscribers globally.  Revenue – from the subscriptions fees its customers pay. Driving forces behind is that customers feel they are getting Value for money. What Problem is AI helping to Solve ?  Consumers are not short of entertainment options today – Streaming movies, Internet, Video games & TV etc.  Earlier TV programs scheduling was precise science –What times to fit in with out lives & earn loyalty.  With on-demand forms of entertainment, this is often no longer possible.  Customers being able to watch what they want, when they want, caused quandary.  What is customers pick up a wrong program to watch & end up not feeling good about value for money .
  23. 23. Netflix- How AI used ? 5 -Use Cases of AI/Data/Machine Learning at Netflix 1. Personalization of Movie Recommendations — Users who watch A are likely to watch B. This is perhaps the most well known feature of a Netflix. Netflix uses the watching history of other users with similar tastes to recommend what you may be most interested in watching next so that you stay engaged and continue your monthly subscription for more. 2. Auto-Generation and Personalization of Thumbnails / Artwork—Using thousands of video frames from an existing movie or show as a starting point for thumbnail generation, Netflix annotates these images then ranks each image in an effort to identify which thumbnails have the highest likelihood of resulting in your click. These calculations are based on what others who are similar to you have clicked on. One finding could be that users who like certain actors / movie genres are more likely to click thumbnails with certain actors/image attributes.
  24. 24. Netflix- How AI used ? 5 -Use Cases of AI/Data/Machine Learning at Netflix ( Continued) 3. Location Scouting for Movie Production (Pre-Production)— Using data to help decide on where and when best to shoot a movie set —  given constraints of scheduling (actor/crew availability), budget(venue, flight/hotel costs), and production scene requirements (day vs night shoot, likelihood of weather event risks in a location). Notice this is more of a data science optimization problem rather than a machine learning model that makes predictions based on past data. 4. Movie Editing (Post-Production) —Using historical data of when quality control checks have failed in the past (when syncing of subtitles to sound/movements were off in the past) — to predict when a manual check is most beneficial in what could otherwise be a very time- intensive and laborious process. 5. Streaming Quality— Using past viewing data to predict bandwidth usage to help Netflix decide when to cache regional servers for faster load times during peak (expected) demand.
  25. 25. Results, Key Challenges, Learning Points & Takeaways Netflix is able to accurately recommend content to viewers based on their preferences & the preferences of others who match their profile. Leads to re-subscribing for longer, offering longer life time value to the company. AI compression algorithms minimizes the size of files that have to be transmitted, improving streaming quality, reduced data usage by a factor 1000. Moving from mail-order to a subscription model vastly increased the amount of data that Netflix was able to collect, not just about what customers watch, but how and when they watch it. Providing accurate recommendations through ongoing dataset of customer habits. Streaming HD and Ultra HD video at lower cost.
  26. 26. Using AI to cover Local News Stories
  27. 27. Netflix  PA is a UK based news agency that provides text & videos news stories, photography, copywriting.  The local news industry in the UK and globally has been in a state of decline for the past decade.  Dwindling audiences, as their readers have switched to the internet and social media for their news, means advertisers now choose to spend their money elsewhere.  This causes problems beyond local news journalists increasingly finding themselves out of work. Who will provide the service of holding local authorities to account while shining a spotlight on issues of local importance?  Well, it turns out, robots might. Leading UK news agency Press Association (PA) is betting that AI can fill the gap left by redundant reporters and shuttered local newspaper offices.  A new initiative sees them partnered with news automation specialists Urbs Media – and endorsed by a 706,000 Euro Google grant – to create 30,000 localized news reports every month.
  28. 28. Technology  The technology relies on natural language generation (NLG), a cornerstone of much of the progress which has been made in recent years thanks to artificial intelligence and automation.  The principle is that humans and computers can work together much more effectively if we all speak the same language. And teaching ultra-fast, infallible machines to understand and communicate in our own human languages is more efficient than teaching slow, fallible humans to communicate with computers in their language (for example, by learning computer code).  It’s the technology behind iPhone’s Siri and Amazon’s Alexa as well as “Chatbots” which are increasingly taking on customer service roles. Only here instead of answering questions their job is to write news stories based on the data it is fed.  The PA project is known as RADAR – Reporters and Data and Robots – and relies on open data sets from government, local authorities and public services. Urbs Media editor-in-chief Gary Rogers told me that they initially started looking at the possibilities of generating stories for national media using open data sources, but soon realized that its highly geographically- segmented nature meant it was very well suited for local stories.
  29. 29. Results, Key Challenges, Learning Points & Takeaways  News stories created with RADAR system are now available to more than 1000 local news outlets through the PA news feed.  Mass scale localization of news stories means there is more chance that important issues will be bought to the public’s attention via local news outlets whose budgets are tight.  AI can quickly and accurately compile news reports in easy to understand natural human language, simply using public datasets.  Human journalists will have more time available to carry out in depth investigation into the background issues that may not be picked up by the AI working on the data alone.
  30. 30. Using AI to find New Music You Will Love
  31. 31. Spotify  Spotify the largest on-demand music service in the world, has a history of pushing technological boundaries and using big data, artificial intelligence and machine learning to drive success.  The digital music company with more than 180 million active users has been busy this year enhancing its service and tech capabilities through several acquisitions. Data: Powerful By-product of Streaming Music  When you have tens of millions of people listening to music every minute of the day, you have access to an extraordinary amount of intel that includes what songs get the most play time, to where listeners are tuning in from and even what device they are using to access the service.  There’s no doubt Spotify is a data- driven company and it uses the data in every part of the organization to drive decisions.  As the service continues to acquire data points, it’s using that information to train the algorithms and machines to listen to music and extrapolate insights that impact its business and the experience of listeners.  One example is the Discover weekly feature on Spotify that reached 40 Million people in its first year.  Every user gets a personalized playlist every week from Spotify of music that they have not heard before on the service, but that will be something the listener is expected to enjoy—a modern-day version of a best friend creating a personalized mix tape.
  32. 32. Spotify  Data for Spotify’s discover weekly playlists is gathered by monitoring its users listening habits through a process of Collaborative Filtering.  Consider Person A regularly listens to music by Artist A and Artist Y. Person B regularly listens to Artist Y and Artist Z.  With this data , a Collaborative filtering algorithm can deduce, with some certainty , that Person A might enjoy being introduced to Artist Z, and Person B might enjoy the output of Artist X.  Audio analysis breaks down each individual track into its constituent parts- for example, tempo, beat, pitch of the notes, types of instruments and sounds used, and the prominence & pattern of lyrics.  Spotify uses deep learning and neural nets to bring all of this information together & make recommendations it knows- to a high degree of probability
  33. 33. Results, Key Challenges, Learning Points & Takeaways  Spotify’s discover weekly playlists mean that it is able to recommend new music that its users will love, and in return they are likely to remain as subscribers to the service.  Big streaming services like Spotify have access to so much data that they can make highly accurate predictions, even about very personal & human issues such as our taste in music.
  34. 34. Using AI to Connect the Unconnected
  35. 35. Telefonica  Telefonica – the global mobile and broadband provider which trades in other parts of Europe as O2 – is heavily invested in the concept and has rolled out a number of initiatives and pilots.  Many of these projects are now in action across major cities in its home nation of Spain.  The concept of a “smart city” is quickly becoming a reality thanks to the ongoing rollout of the Internet of Things (IoT) – the ever-growing number devices in our personal and business lives which are online, connected and capable of sharing data.  The term “smart city” refers to the segment of the IoT which concerns running our civic and municipal infrastructure and amenities, from road and transport systems to waste collection and power distribution.  Citizen access to democratic processes such as decision-making, open data and elections can also be improved with smart technology.
  36. 36. Telefonica  “The importance is that this shows how a city can move from being an efficient city to a truly intelligent city - these are great examples of how smart cities are starting to evolve here in Spain and hopefully Europe.”  In waste management, sensors attached to all refuse containers report how close to capacity they are in real-time, allowing for far more efficient allocation of resources to emptying them.  It also means KPIs can be more closely tied to bottom-line impact – how many bins are close to overflowing and won’t be emptied within the next few hours? – rather than abstracted variables which could be monitored previously, such as the number of waste collection trucks on the road.  “This is changing the way that services in these cities are being contracted”, Garcia Gomez tells me. “The service provider has 20 KPIs that they have to meet to get their bonus.”
  37. 37. Telefonica  Playing to another of the strengths of smart city initiatives, different applications can be created to let different groups of people take advantage of the tools and data available.  In this case, citizens can make use of a smartphone app which lets them tag overflowing bins in their neighborhood, or bring offensive graffiti to the attention of contracted cleaners.  Targets can be set for how quickly contractors are expected to rectify the situation and claim their fee.  The sensor data is also combined with population and demographic data from open sources – “So you can see, for example, how many people over the age of 60 are affected by overflows.  Is an 80-year-old going to walk an extra kilometer because her container is full? Probably not. So that gives the city greater info to better serve its citizens,” says Garcia Gomez.  Car parking in a smart city – such as Valencia, one of the locations of Telefonica’s pilots – is monitored through the use of sensors monitoring parking bays. This means that information is available on the density of parking across the city in real time. This helps set parking fees so a more even distribution of parking can be achieved, reducing congestion as well as pollution.
  38. 38. Results, Key Challenges, Learning Points & Takeaways  The Machine learning & Computer vision component of the program was able to map 95% of the population of the remote areas that were analyzed, with a less than 3% rate of false positives.  Isolated populations have not benefited from the breakthroughs in communication technology enabled by the internet across the developed world.  AI makes it possible to map population densities using satellite images to give more accurate data on where people are living.  Analysis of transport infrastructure enables technology to be deployed in the most cost- effective way, making it viable to connect previously unreachable populations to the internet.
  39. 39. Using AI to fight fake News & Spambots
  40. 40. Twitter  Twitter is a social media platform where 328 million monthly active users microblog (share 280-character updates) with their followers.  It’s a cross between instant messaging and blogging—or social messaging, but it’s also been crucial for news reporting, event promotion, marketing and business.  Whether you’re a fan or not, Twitter has become the 9th largest social network in the world and Cortex, Twitter’s in-house engineering team, has turned to the power of artificial intelligence (AI) to help enhance the platform’s user experience. How Twitter uses AI in practice  Billionaire Mark Cuban said in 2017 he purchased Twitter stock because “they finally got their act together with artificial intelligence.”  In 2016, Twitter purchased Magic Pony Technology to bulk up Cortex, who wants to “build the most advanced AI platform in the world, at Twitter scale, to apply the most complex AI algorithms to our most challenging datasets, seamlessly.”
  41. 41. Twitter  One of the ways Twitter uses artificial intelligence is to determine what tweet recommendations to suggest on users’ timelines with the goal of highlighting the most relevant tweets for every individual.  Prior to this shift, Twitter would show its users tweets in reverse chronological order.  Today, the algorithms scan and score thousands of tweets per second to rank them for every user’s feed.  The social media platform also deployed AI to fight inappropriate and racist content on its platform as authorities in the UK and Germany increase measures and fines to prevent hate speech, fake news and illegal content on social media. In the first six months of 2017, Twitter took down nearly 300,000 terrorist accounts which were identified by its AI tools; in fact, 95 percent of suspended terrorism-related accounts were identified by algorithms rather than human users.  Artificial intelligence tools are also supporting small tweaks to Twitter that improve the overall user experience. For example, its image Cropping tools are using AI to automatically crop images in a much more appealing way and are monitoring live video feeds and categorizing them based on subject matter to improve their searchability and to help the algorithms identify videos users might be interested in seeing in their feeds.
  42. 42. Twitter Technical details  Twitter’s ranking algorithm has taken in lots of data, processed it through deep neural networks and learned over time what content would be relevant for each individual user.  All tweets are scored on a ranking model that is used to determine the probability if a user would value that content in their feed.  The ranking model considers the content of the tweet itself including if it is accompanied with an image or video and how many retweets or likes it has received; the author of the tweet to see if you had any past interactions with the author and the strength of your connection to the author; and considers the type and tone of tweets you have a history of liking in the past and how this tweet resembles others that you seemed to appreciate.  The higher the relevancy score, the higher on your feed you will see the tweet and the probability the tweet will appear in the “In case you missed it” module.  As Twitter continues to refine its algorithms, it has to balance the speed as well as quality of the algorithm’s review process to meet the requirements of the platform for real-time updates. In addition to the speed and quality of predictions, algorithms are assessed on their resource demands and maintainability over time.
  43. 43. Twitter  The company uses IBM Watson & its natural language processing skill-set to track and remove abusive messages since the AI tech of Watson not only understands natural language but can infer intonation and extract meaning from images quickly—it can analyze millions of tweets in a second.  Twitter’s AI tools “crop using saliency” to show the most interesting aspect of images whether they are faces or not.  In order to train the tool, the Twitter team defined what is most salient by using data from academics that studied eye tracking.  Then, in order to optimize the tool in real time on the site, they used AI to train a quicker version of the tool to speed up the crops.  First, they trained a smaller network by using the first program that was good, but slow. Then, the software engineers streamlined the process by removing the less important visual cues on images.  In order to train its algorithm to recognize what’s happening on a live feed, Cortex used deep learning. They trained a large neural network to recognize the content on video from a large number of examples.  Humans watched videos and tagged them with several keywords to identify what they saw. So, a video of a dog wasn’t just tagged with the keyword dog, but also animal, canine, mammal and more. That data was then used to train the algorithm so it could then identify content in video.
  44. 44. Results, Key Challenges, Learning Points & Takeaways In two months, Twitter used automated detection tools to take down more than 70 million “ fake & suspicious” accounts. Year on year 214% accounts have been removed for violating span policies. Behavior of Scammers & those with malicious intent can be identified by their behavior online using AI. Twitter realizes that the freedom of speech its platform offers is important, but has put its belief that the safety of users is paramount at the top of its agenda.
  45. 45. Using Machine Learning to assess service quality
  46. 46. Verizon  Verizon’s FIOS fiber optic broadband keeps millions of US homes online. However, monitoring stability and reacting to faults and outages which affect customer experience takes huge amounts of resources.  Until recently, Verizon primarily relied on customer feedback to understand when the speed and quality of its service was falling short of expectations.  In recent years, however, following a large investment in analytics and AI-driven technology such as Machine Learning – in part subsumed through the company’s 2017 acquisition of Yahoo! and it’s research units – a different approach is bringing impressive results.  Now it’s predictive analytics algorithms monitor 3GB of data every second streaming from millions of network interfaces – from customers’ routers to an array of sensors gathering temperature and weather data, and software which “listens in” on operational data, such as billing records.  Verizon’s analytics infrastructure allowed them to predict 200 “customer impacting” events before they happened and take steps to prevent them occurring.  “Essentially what they are trying to do is listen to all of their network elements … there’s a tremendous wealth of data that comes from the different elements and they want to listen to them, translate them, run them through [predictive] models and ensure that there’re no interruptions to their customers.”
  47. 47. Verizon  Verizon’s AI and Big Data infrastructure is built largely from open source components. The team heavily relies on Spark and Kafka due to their ability to handle very fast streaming network data in real-time.  “If you think about it, it makes sense,” says Tegerdine, “the data never stops flowing so we need real time processing to respond to it.”  The platform sits on Hadoop, and development work is carried out in Python and Java.  Another core strategy has been the deployment of “incubation teams”, comprised of specialists in different areas of data technology. These are comprised of data scientists, data engineers, data architects and, crucially, a data translator.  The data translators are a very powerful and unique layer. They can speak the data science language but they also know the business – typically these are people we have pulled out of business functions.  Data scientists speak a particular language but data translators make it real. They’re the glue that ties it all together.”  As far as the future goes, AI (and machine learning in particular) will play an increasingly vital role in protecting and ensuring performance, and networks become bigger, faster and more complex.
  48. 48. Results, Key Challenges, Learning Points & Takeaways Noticeable increase in sales through extensive monitoring & testing of the service, engineers surprised to learn that 750MB /s service was actually consistently providing speeds of 1 GB/s into customer homes- Ability to rebrand. Predicting customer-impacting events means that they can be fixed before they cause issues, leading to higher customer satisfaction & reduced churn rates. The size & reach of Verizon’s network mean that there is a wealth of data available that can be used to predict these events. The fact that much of Verizon’s data is internal data & not available to other businesses gives it a competitive advantage.
  49. 49. Using AI to Stream Videos Faster & Improve Customer experience
  50. 50. Viacom  Viacom, owner of household brands such as Comedy Central, Nickelodeon and MTV, is one of the largest media companies in the world, delivering more than 170 cable, broadcast and online networks in around 160 countries.  Monitoring of the digital networks which are used to pump their content into millions of homes gives them access to a huge amount of data, on how both their systems and their audiences behave.  Like many large organizations, Viacom is currently manoeuvring to position itself as a fully data-driven company – with as few of the important decisions left to guess-work and gut-feeling as possible.  In fact, Viacom is so heavily involved in Big Data, that as well as luring talent from tech companies such as Microsoft in recent years, it has even produced a Super-Cool rap Video to proclaim its love.  As part of this drive, it has developed a number of use cases for the network data it is collecting. Dan Morris, senior director of product analytics, took the time to talk to me about two of them – explaining how they combined this network data with cutting edge real time analytics, to improve viewer experience and increase customer retention.
  51. 51. Viacom  The Strategy-Customer Focus from the ground up – with reducing customer dissatisfaction through poor service as the problem to be overcome.  Through Machine learning algorithms to firstly establish the “normal” behaviors that are expected on the network. Then it identifies “outlier” data which sits outside this threshold of normal behavior, and attempts to recognize events that have led to the emergence of these outliers.  “The beauty of this is that they don’t just look at one singular data source like interface statistics – They also going out and collecting things like environmental statistics, CPU usage on routers … They use machine learning to learn what ‘normal’ is.  The strategy will become increasingly important as it moves towards the goal of deploying the first residential and mobile 5G networks at the end of this year.  On its home networks, Verizon runs automated testing on a sample of 60,000 in-home routers every two hours, to ensure that customers are receiving the speed of service they are paying for.
  52. 52. Viacom Improving video experience  Streaming petabytes of video data across the world puts a strain on the delivery systems – we’re all familiar with online videos either failing to load or constantly stuttering as they rebuffer.  At the same time there are areas where excess bandwidth is going to waste – when everyone is asleep or out at work. Viacom has built a real-time analytics platform based around Apache Spark and Databricks, which constantly monitors the quality of video feeds and reallocates resources in real-time when it thinks it will be needed.  “Delivery of video is at the core of everything they do, and their goal is to be exceptional at that. But there’re a lot of variables at play – they have internal systems talking to external systems, they have content delivery, they have ad severs, and on the user side there’s a whole bunch of environmental factors like Wifi connectivity which they really have no control over.”  While they are out of Viacom’s control, monitoring these factors can potentially lead to insights which could be used to predict when problems will emerge. In other words while they may not be able to choose whether a viewer will watch on TV, or on a tablet, their choice can be anticipated and planned for.  “So, with what we’ve built on the data side, we’re able to take all this raw data and do two things – surface trends, and isolate areas of opportunity to present a superior viewing experience.”
  53. 53. Viacom  Two metrics are particularly closely monitored for this process – “time to first frame” (how long it takes a video to start) and re-buffering rate – how often the video stutters.  These have been shown by the analytics to be primary indicators of whether an audience will settle in to watch a video, or search for something else instead.  Thanks to improvements driven by their analytics, Viacom has reduced the time-to- first-frame their viewers experience to around one third of what it was, across their network.  “Particularly with younger audiences, they know this is super-important. That split second could be the difference between retaining the user and losing them. So, that was definitely a huge area of focus and the data was an instrumental part.” Growing the audience  For the second use case, Morris’s team again found their “north star” metric – the insight that could guide them to their goal. While Facebook acknowledges its North Star is to get new users to seven friends within ten days, at which point they are hooked, Viacom deduced, through analytics, that it needs to get viewers hooked on at least two individual shows.  Morris tells me “What we found is that if we can get you to watch two shows you’re 350% more likely to stay with us. If we can get you to watch four shows, that goes up to 700% - seven times more likely.”
  54. 54. Looking at Future -Viacom Big Data is certainly set to play a big part in Viacom’s immediate future. Now its usefulness has been proven, with concrete results, the next step is to introduce it across the business, wherever a use can be found. “Now it’s about – we’ve had a lot of success, how can we share this with other teams?” Morris says. “The value we’ve seen within our team is that it lets us focus on business problems and makes certain processes very simple. Now it’s a question of how do we bring these benefits to others in the organization who might not be aware of what they can do with this type of platform. “The truth is, there’re endless opportunities – we’re looking at cognitive computing and a whole bunch of different use cases – we really want to embed data into every single process throughout our company.”
  55. 55. Results, Key Challenges, Learning Points & Takeaways Viacom’s social media analytics platform helps it to measure the impact posting different content at different times of the day & on different networks has on customer viewing habits. Distributing resources at its disposal for promoting content. Reduction in streaming time by 33% over its web services. Social media offers unparalleled opportunity to get to know the customer; but may need advanced tools such as AI to cut through all of the noise & find the insights that matter. AI is now powerful enough to implement real-time monitoring & automate resource management across vast data networks. Identifying key drivers of success – a North pole metric – is a primary use case for many AI analytics initiatives in business.
  56. 56. Mail your comments to ramaddster@gmail.com End of Part -3 Will continue the summary in Part - 4