One of the earliest machine learning use cases for G Suite was within Gmail. Historically, Gmail used a rule-based system, which meant our anti-spam team would create new rules to match individual spam patterns. Over a decade of using this process, we improved spam detection accuracy to 99%. Starting in 2014, our team augmented this rule-based system to generate rules using machine learning algorithms instead, taking spam detection one step further. Now, we use Tensor Flow and other machine learning to continually regenerate the "spam filter," so the system has learned to predict which emails are most likely junk.

Piracy has always been a problem for content like online video, but Facebook looks to have an unconventional pitch to content owners who've had their videos reuploaded: You'll still be able to make money on them. In a post Thursday, Facebook announced a tweak to how it processes videos that are pirated and reuploaded from one owner to another. When Facebook's rights management system detects a video that has been pirated, the original owner can now choose to receive ad earnings from the duplicated clip. With Rights Manager, rights owners can find matches of their video content on Facebook; these matches are surfaced on a dashboard. Previously, the rights owner would review these matches in the dashboard to take action.

Building a chatbot is really about taking computer-human conversation to a whole new level. Technology experts generally talk about two methods of building chatbots. The first is a rule-based approach, where the developer writes rules for the system, or in other words, employs hard coding in building the chatbot. The second method entails the use of machine learning, where a massive amount of streaming data is used, and the system learns on its own. AIM caught up with Aditya Chavan, Head of Marketing, and Shashank Prasad, Head of Infrastructure, representing Machaao.

Email marketing is in for a complete overhaul. With AI, email marketing isn't limited to rule-based triggers but has evolved into a means of more in-the-moment personalization. AI has filled in the missing gap between traditional shopping and online shopping with 1:1 personalization. Here are four of the ways AI has changed email marketing for better. You have three seconds to seize the shopper's attention.

It seems like artificial intelligence is everywhere. No longer the stuff of Ridley Scott and Stanley Kubrick flicks, AI has rapidly wormed its way into everyday news coverage and real-world business conversations. Since last April alone, the amount of published articles, blog posts and multimedia content featuring the words "AI" or "Artificial Intelligence" has more than doubled, according to Factiva. Talk of AI often centers around life-altering technological advancements such as driverless vehicles or genomic medicine. But the ad and marketing tech industry, always willing to capitalize on a trend, has joined in with a flood of new digital ad and marketing platforms and services branded as AI-fueled technologies.

Securifi's Almond 3 has two features you won't find in competing whole-home Wi-Fi systems: a touchscreen and a built-in smart home hub. Where other manufacturers take pride in hiding most of the inner workings of their user-friendly routers in the name of ease-of-use, Securifi gives the router enthusiast full access to all its levers and dials. But you don't need to be an enthusiast to appreciate the Almond 3--it's super easy to set up using Securifi's smartphone app, web interface, the router's 2.8-inch touchscreen, or any combination of the three. You'll find a ZigBee HA 1.2 radio onboard for use with sensors (door/window, motion, flood, smoke, and other types), lighting controls, thermostats, smart entry locks, and more. Securifi makes some of its own smart home devices, including a petite smart plug and door/window, motion, and flood sensors, but just about any ZigBee device should be compatible. Securifi lists specific compatible products on its website, a list that includes Philips Hue lighting, Nest thermostats and smoke detectors, and Yale smart entry locks.

Formal concepts and closed itemsets proved to be of big importance for knowledge discovery, both as a tool for concise representation of association rules and a tool for clustering and constructing domain taxonomies and ontologies. Exponential explosion makes it difficult to consider the whole concept lattice arising from data, one needs to select most useful and interesting concepts. In this paper interestingness measures of concepts are considered and compared with respect to various aspects, such as efficiency of computation and applicability to noisy data and performing ranking correlation.

TokensRegex is a generic framework included in Stanford CoreNLP for defining patterns over text (sequences of tokens) and mapping it to semantic objects represented as Java objects. TokensRegex emphasizes describing text as a sequence of tokens (words, punctuation marks, etc.), which may have additional attributes, and writing patterns over those tokens, rather than working at the character level, as with standard regular expression packages. TokensRegex was used to develop SUTime, a rule-based temporal tagger for recognizing and normalizing temporal expressions. An included set of slides provides an overview of this package. There is quite detailed Javadoc for several of the key classes: for the matching patterns, see the Javadoc for TokenSequencePattern and for actions, see the Javadoc for Expressions.

Search Beyond is a forum that brings together the latest discussion from some of the UK's most innovative independent digital agencies. A handful of senior practitioners meet on a quarterly basis to debate a topic facing the digital marketing industry, with the insights and output appearing here on Think with Google. In our second Search Beyond session, we wanted to hear how independent agencies are preparing to face down increasing complexities in the digital landscape by adopting machine learning. The participants launched into the discussion by revealing how they're already putting these tools to use in their own work on behalf of clients today. "An element of automation is machine learning, so that is a lot of what we do", said Maria Yiangou, Group Account Director at All Response Media.

We present an algorithm for building probabilistic rule lists that is two orders of magnitude faster than previous work. Rule list algorithms are competitors for decision tree algorithms. They are associative classifiers, in that they are built from pre-mined association rules. They have a logical structure that is a sequence of IF-THEN rules, identical to a decision list or one-sided decision tree. Instead of using greedy splitting and pruning like decision tree algorithms, we fully optimize over rule lists, striking a practical balance between accuracy, interpretability, and computational speed. The algorithm presented here uses a mixture of theoretical bounds (tight enough to have practical implications as a screening or bounding procedure), computational reuse, and highly tuned language libraries to achieve computational efficiency. Currently, for many practical problems, this method achieves better accuracy and sparsity than decision trees; further, in many cases, the computational time is practical and often less than that of decision trees. The result is a probabilistic classifier (which estimates P(y = 1|x) for each x) that optimizes the posterior of a Bayesian hierarchical model over rule lists.