This paper describes a practical system for Multi Touch Attribution (MTA) for use by a publisher of digital ads. We developed this system for JD.com, an eCommerce company, which is also a publisher of digital ads in China. The approach has two steps. The first step ('response modeling') fits a user-level model for purchase of a product as a function of the user's exposure to ads. The second ('credit allocation') uses the fitted model to allocate the incremental part of the observed purchase due to advertising, to the ads the user is exposed to over the previous T days. To implement step one, we train a Recurrent Neural Network (RNN) on user-level conversion and exposure data. The RNN has the advantage of flexibly handling the sequential dependence in the data in a semi-parametric way. The specific RNN formulation we implement captures the impact of advertising intensity, timing, competition, and user-heterogeneity, which are known to be relevant to ad-response. To implement step two, we compute Shapley Values, which have the advantage of having axiomatic foundations and satisfying fairness considerations. The specific formulation of the Shapley Value we implement respects incrementality by allocating the overall incremental improvement in conversion to the exposed ads, while handling the sequence-dependence of exposures on the observed outcomes. The system is under production at JD.com, and scales to handle the high dimensionality of the problem on the platform (attribution of the orders of about 300M users, for roughly 160K brands, across 200+ ad-types, served about 80B ad-impressions over a typical 15-day period).

Sketching refers to a class of randomized dimensionality reduction methods that aim to preserve relevant information in large-scale datasets. They have efficient memory requirements and typically require just a single pass over the dataset. Efficient sketching methods have been derived for vector and matrix-valued datasets. When the datasets are higher-order tensors, a naive approach is to flatten the tensors into vectors or matrices and then sketch them. However, this is inefficient since it ignores the multi-dimensional nature of tensors. In this paper, we propose a novel multi-dimensional tensor sketch (MTS) that preserves higher order data structures while reducing dimensionality. We build this as an extension to the popular count sketch (CS) and show that it yields an unbiased estimator of the original tensor. We demonstrate significant advantages in compression ratios when the original data has decomposable tensor representations such as the Tucker, CP, tensor train or Kronecker product forms. We apply MTS to tensorized neural networks where we replace fully connected layers with tensor operations. We achieve nearly state of art accuracy with significant compression on image classification benchmarks.

We present a new method for design problems wherein the goal is to maximize or specify the value of one or more properties of interest. For example, in protein design, one may wish to find the protein sequence that maximizes fluorescence. We assume access to one or more, potentially black box, stochastic "oracle" predictive functions, each of which maps from input (e.g., protein sequences) design space to a distribution over a property of interest (e.g. protein fluorescence). At first glance, this problem can be framed as one of optimizing the oracle(s) with respect to the input. However, many state-of-the-art predictive models, such as neural networks, are known to suffer from pathologies, especially for data far from the training distribution. Thus we need to modulate the optimization of the oracle inputs with prior knowledge about what makes `realistic' inputs (e.g., proteins that stably fold). Herein, we propose a new method to solve this problem, Conditioning by Adaptive Sampling, which yields state-of-the-art results on a protein fluorescence problem, as compared to other recently published approaches. Formally, our method achieves its success by using model-based adaptive sampling to estimate the conditional distribution of the input sequences given the desired properties.

We study a natural variant of the implicational fragment of propositional logic. Its formulas are pairs of conjunctions of positive literals, related together by an implicational-like connective; the semantics of this sort of implication is defined in terms of a threshold on a conditional probability of the consequent, given the antecedent: we are dealing with what the data analysis community calls confidence of partial implications or association rules. Existing studies of redundancy among these partial implications have characterized so far only entailment from one premise and entailment from two premises, both in the stand-alone case and in the case of presence of additional classical implications (this is what we call "relative entailment"). By exploiting a previously noted alternative view of the entailment in terms of linear programming duality, we characterize exactly the cases of entailment from arbitrary numbers of premises, again both in the stand-alone case and in the case of presence of additional classical implications. As a result, we obtain decision algorithms of better complexity; additionally, for each potential case of entailment, we identify a critical confidence threshold and show that it is, actually, intrinsic to each set of premises and antecedent of the conclusion.

Many real-world reinforcement learning tasks require multiple agents to make sequential decisions under the agents' interaction, where well-coordinated actions among the agents are crucial to achieve the target goal better at these tasks. One way to accelerate the coordination effect is to enable multiple agents to communicate with each other in a distributed manner and behave as a group. In this paper, we study a practical scenario when (i) the communication bandwidth is limited and (ii) the agents share the communication medium so that only a restricted number of agents are able to simultaneously use the medium, as in the state-of-the-art wireless networking standards. This calls for a certain form of communication scheduling. In that regard, we propose a multi-agent deep reinforcement learning framework, called SchedNet, in which agents learn how to schedule themselves, how to encode the messages, and how to select actions based on received messages. SchedNet is capable of deciding which agents should be entitled to broadcasting their (encoded) messages, by learning the importance of each agent's partially observed information. We evaluate SchedNet against multiple baselines under two different applications, namely, cooperative communication and navigation, and predator-prey. Our experiments show a non-negligible performance gap between SchedNet and other mechanisms such as the ones without communication and with vanilla scheduling methods, e.g., round robin, ranging from 32% to 43%.

Despite recent setbacks, solar remains a bright spot in the often wobbly renewable energy sector. In the U.S., the solar market is projected to top $22.90 billion by 2025, driven by falling materials costs and growing interest in offsite and rooftop installations. Moreover, in China -- the world's leading installer of solar panels and the largest producer of photovoltaic power -- 1.84 percent of the total electricity generated in the country two years ago came from solar. There's clearly growth -- which San Francisco startup Aurora Solar seeks to capitalize on with a novel solution combining lidar data, computer-assisted design, and computer vision. The company, which develops a suite of software that streamlines the solar panel installation process, today announced it has secured $20 million in a Series A round of financing from Energize Ventures, with contributions from S28 Capital and existing investor Pear.

The Land of the Rising Sun is a peculiar mix of tradition and modernity. Nowhere else in the world can one see centuries-old shrines sitting comfortably next to high tech skyscrapers in such harmony. Nowhere in the world does the airport ground crew stop what they're doing so they can wave goodbye to departing planes until they're out of sight. Nowhere in the world will you find customer service that drips with genuine sweetness with no tips expected. Nowhere in the world will you find a people as endearing as the Japanese.

When it comes to digital transformation, a term that has almost become synonymous with it is Artificial Intelligence. The technology has made an impact on almost every aspect of every business. The scale of importance artificial intelligence holds in the current space can be gauged by the fact that even the Government of India announced special initiatives for it in the Union Budget 2019. A National Programme on'Artificial Intelligence' has been envisioned by the government, which would be catalysed by the establishment of the National Centre on Artificial Intelligence as a hub along with Centres of Excellence. Also, nine priority areas have been identified for the same and a National Artificial Intelligence portal will be developed soon.

Medical diagnostic services using artificial intelligence are set to be introduced this year in Japan, where the adoption of information technology in health care has been slow, sources said Monday. The private Showa University in Tokyo together with Nagoya University and Cybernet Systems Co. have jointly developed software that analyzes colon polyps shown in images taken during endoscopic examinations. Using a vast amount of past diagnostic data, the software can determine whether such polyps are malignant. It has been proven reliable -- with an assessment determining that it can identify a potentially cancerous polyp with the same accuracy as a leading specialist -- and approved for commercialization by the government. LPixel Inc., an image analysis service venture, has produced a program that spots cell degeneration in the cerebrum by examining images taken during a magnetic resonance imaging (MRI) scan.

After a couple of years spent testing its autonomous-drive system in Moscow and elsewhere in Russia, Yandex -- the Russian equivalent of Google -- showcased its self-driving product at CES in January. The results were encouraging, based on my 20-minute drive in downtown Las Vegas traffic in a Toyota Prius fitted with the Yandex equipment. The ride was smooth -- actually, smoother than other test cars I have tried in the past months. I would say the vehicle's behavior had a much more "human" feeling than I had expected, with smooth turns, less hesitation when merging into traffic, good acceleration and no sudden braking. The route included unprotected left-hand turns, pedestrian crossings and busy traffic with speeds exceeding 70 kph.