Learning useful representations is a key ingredient to the success of modern machine learning. Currently, representation learning mostly relies on embedding data into Euclidean space. However, recent work has shown that data in some domains is better modeled by non-euclidean metric spaces, and inappropriate geometry can result in inferior performance. In this paper, we aim to eliminate the inductive bias imposed by the embedding space geometry. Namely, we propose to map data into more general non-vector metric spaces: a weighted graph with a shortest path distance. By design, such graphs can model arbitrary geometry with a proper configuration of edges and weights. Our main contribution is PRODIGE: a method that learns a weighted graph representation of data end-to-end by gradient descent. Greater generality and fewer model assumptions make PRODIGE more powerful than existing embedding-based approaches. We confirm the superiority of our method via extensive experiments on a wide range of tasks, including classification, compression, and collaborative filtering.

In recommender systems, the user-item interaction data is usually sparse and not sufficient for learning comprehensive user/item representations for recommendation. To address this problem, we propose a novel dual-bridging recommendation model (DBRec). DBRec performs latent user/item group discovery simultaneously with collaborative filtering, and interacts group information with users/items for bridging similar users/items. Therefore, a user's preference over an unobserved item, in DBRec, can be bridged by the users within the same group who have rated the item, or the user-rated items that share the same group with the unobserved item. In addition, we propose to jointly learn user-user group (item-item group) hierarchies, so that we can effectively discover latent groups and learn compact user/item representations. We jointly integrate collaborative filtering, latent group discovering and hierarchical modelling into a unified framework, so that all the model parameters can be learned toward the optimization of the objective function. We validate the effectiveness of the proposed model with two real datasets, and demonstrate its advantage over the state-of-the-art recommendation models with extensive experiments.

Recent years have witnessed the great success of deep neural networks in many research areas. The fundamental idea behind the design of most neural networks is to learn similarity patterns from data for prediction and inference, which lacks the ability of logical reasoning. However, the concrete ability of logical reasoning is critical to many theoretical and practical problems. In this paper, we propose Neural Logic Network (NLN), which is a dynamic neural architecture that builds the computational graph according to input logical expressions. It learns basic logical operations as neural modules, and conducts propositional logical reasoning through the network for inference. Experiments on simulated data show that NLN achieves significant performance on solving logical equations. Further experiments on real-world data show that NLN significantly outperforms state-of-the-art models on collaborative filtering and personalized recommendation tasks.

Google unveiled a new Pixel smartphone and other hardware devices Tuesday, all aimed at getting people even more dependent on its artificial-intelligence services. The Pixel 4 phone promises to respond to AI queries even faster than before, while a home Wi-Fi system is getting the AI features for the first time. The company also unveiled a new smart speaker and wireless earbuds, both invoking the AI-powered Google Assistant. The Assistant, akin to Apple's Siri and Amazon's Alexa, is now available on more than 1 billion devices, including ones made by other manufacturers. With Google's own products, though, the company can steer users to Assistant features even more.

Google designers show their latest devices during a visit to its hardware studio where they designed the company's new Pixel 4 flagship phone, Pixel Go laptop, Pixel Buds wireless headphones, Nest Mini smart speaker and Nest Wifi system. Though the Pixel 4 and Pixel 4 XL smartphones may be the stars of the Made by Google media event in New York City, a new and impressive smart speaker garnered quite a bit of chatter, too. Nest Mini ($49), which is about the size of a doughnut, has some improved hardware and software to go up against its primary competitor, Amazon Echo Dot. This diminutive smart speaker offers twice the bass as the original Google Home Mini, the company says. Indeed, the audio sounded fuller and louder than its predecessor when we cranked it up in a post-event demo session.

Today at its Google hardware event, Google introduced new mesh routers called Nest Wifi. This is a successor tot he Google Wifi product it introduced a couple of years ago, but with a number of improvements. The new Nest Wifi consists of two types of devices, one a router that plugs into your modem, and one'point' amplifies the signal and extends the network, and it's more powerful so you only need these two things. It's available to pre-order, and will ship on November 4. It comes in a 2-pack or a 3-pack variant, for $249 or $349 respectively, and will be available in eight countries at launch. Google says that Nest Wifi offers 2x better speed than Google Wifi, with up to 25% better coverage.

In this Tuesday, Sept. 24, 2019, photo New Pixel 4 phones are displayed at Google in Mountain View, Calif. SAN FRANCISCO (AP) -- Google unveiled a new Pixel smartphone and other hardware devices Tuesday, all aimed at getting people even more dependent on its artificial-intelligence services. The Pixel 4 phone promises to respond to AI queries even faster than before, while a home Wi-Fi system is getting the AI features for the first time. The company also unveiled a new smart speaker and wireless earbuds, both invoking the AI-powered Google Assistant. The Assistant, akin to Apple's Siri and Amazon's Alexa, is now available on more than 1 billion devices, including ones made by other manufacturers.

Fledgling gal-bots are the latest hires in the virtual assistant landscape. Meet Mia and Marge: two virtual assistants in the banking world – each brought into existence by women, both of whom carry deep institutional knowledge, subject matter expertise and long-standing credibility. UBank's Lee Hatton (Mia) and The Royal Bank of Scotland's (RBS) MaryAnn Fleming (Marge) are among 40 women who have been recognized as 2019's women leaders in A.I. by IBM. These leaders have succeeded in garnering acceptance of A.I. in the workplace, elevating their customers' experience and their companies' brands. It seems mortgage consumer complaints consistently surface around the loan application process according to UBank CEO, Lee Hatton.

The total installed base of devices with Artificial Intelligence (AI) will grow from 2.7 billion in 2019 to 4.5 billion in 2024, forecasts global tech market advisory firm, ABI Research. There are billions of petabytes of data flowing through these AI devices every day; the challenge now facing both technology companies and implementers is getting all these devices to learn, think, and work together. According to a recent whitepaper from ABI Research, Artificial Intelligence Meets Business Intelligence, multimodal learning is the key to making this happen, and it's fast becoming one of the most exciting -- and potentially transformative -- fields of artificial intelligence. "Multimodal learning consolidates disconnected, heterogeneous data from various sensors and data inputs into a single model," says ABI Research chief research officer Stuart Carlaw. "Learning-based methods that combine signals from different modalities can generate more robust inference, or even new insights, which would be impossible in a unimodal system."

Effective and efficient recommendation is crucial for modern e-commerce platforms. It consists of two indispensable components named Click-Through Rate (CTR) prediction and Conversion Rate (CVR) prediction, where the latter is an essential factor contributing to the final purchasing volume. Existing methods specifically predict CVR using the clicked and purchased samples, which has limited performance affected by the well-known sample selection bias and data sparsity issues. To address these issues, we propose a novel deep CVR prediction method by considering the post-click behaviors. After grouping deterministic actions together, we construct a novel sequential path, which elaborately depicts the post-click behaviors of users. Based on the path, we define the CVR and several related probabilities including CTR, etc., and devise a deep neural network with multiple targets involved accordingly. It takes advantage of the abundant samples with deterministic labels derived from the post-click actions, leading to a significant improvement of CVR prediction. Extensive experiments on both offline and online settings demonstrate its superiority over representative state-of-the-art methods.