Conventional solutions to automatic related work summarization rely heavily on human-engineered features. In this paper, we develop a neural data-driven summarizer by leveraging the seq2seq paradigm, in which a joint context-driven attention mechanism is proposed to measure the contextual relevance within full texts and a heterogeneous bibliography graph simultaneously. Our motivation is to maintain the topic coherency between a related work section and its target document, where both the textual and graphic contexts play a big role in characterizing the relationship among scientific publications accurately. Experimental results on a large dataset show that our approach achieves a considerable improvement over a typical seq2seq summarizer and five classical summarization baselines.

Due to the ability of deep neural nets to learn rich representations, recent advances in unsupervised domain adaptation have focused on learning domain-invariant features that achieve a small error on the source domain. The hope is that the learnt representation, together with the hypothesis learnt from the source domain, can generalize to the target domain. In this paper, we first construct a simple counterexample showing that, contrary to common belief, the above conditions are not sufficient to guarantee successful domain adaptation. In particular, the counterexample (Fig. 1) exhibits \emph{conditional shift}: the class-conditional distributions of input features change between source and target domains. To give a sufficient condition for domain adaptation, we propose a natural and interpretable generalization upper bound that explicitly takes into account the aforementioned shift. Moreover, we shed new light on the problem by proving an information-theoretic lower bound on the joint error of \emph{any} domain adaptation method that attempts to learn invariant representations. Our result characterizes a fundamental tradeoff between learning invariant representations and achieving small joint error on both domains when the marginal label distributions differ from source to target. Finally, we conduct experiments on real-world datasets that corroborate our theoretical findings. We believe these insights are helpful in guiding the future design of domain adaptation and representation learning algorithms.

DAVOS, SWITZERLAND – The star who stole the stage at the annual meeting of the World Economic Forum, which just ended here yesterday, wasn't the stuff of flesh and blood but of data-driven algorithms. US President Donald Trump, China's Xi Jinping, India's Narendra Modi, France's Emmanuel Macron and Great Britain's Theresa May were no shows at this gathering of global movers and shakers, occupied with more pressing matters at home. That left hundreds of global business executives with less distraction as they turned their attention to Artificial Intelligence (AI), a term few of them knew even a couple of years ago and a technology they still don't fully comprehend. Yet in one session after another, they shared what they were (or weren't) doing about it and learned how AI would transform their industries, their societies and international relations, perhaps as no technology before it. Not even news late in the week from Venezuela shifted the conversation all that much.

Australian Open champion Naomi Osaka says she had to be a "robot" and turn off her feelings to hold her nerve and win the final against Petra Kvitova. The Japanese, 21, had tears in her eyes after having three match points saved by her Czech opponent in the second set - before winning 7-6 (7-2) 5-7 6-4. "You know how some people get worked up about things? That's a very human thing to do," said Osaka. "Sometimes I feel like I don't want to waste my energy doing stuff like that."

An American man who murdered and beheaded a Japanese woman he met on an online dating app has been sentenced to eight years in prison, according to reports. Yevgeniy Vasilievich Bayraktar, of Long Island, New York, was sentenced on Tuesday in a Japanese court after reportedly admitting to strangling Saki Kondo, 27, in February 2018 at an apartment he'd rented while sightseeing in Osaka. Bayraktar strangled the young woman and then used a saw to dismember her, and buried her body parts across different cities, the New York Post reports. Bayraktar, 27, was not indicted for murder in the case, because prosecutors could not prove that he had set out with the intention to kill Kondo. He was instead charged and found guilty of manslaughter and abandoning a body.

The eleventh edition of the annual Mobile India conference displayed a wide range of insights on the impacts of AI and emerging tech. When was the last time startups got to engage a room full of PHDs and technical domain experts? Well, the Mobile India conference offered them just that and much more - including more papers, demos, posters, and Graduate Student forum presentations. The discussions and deliberations during this year's Mobile India conference revolved around core issues surrounding artificial intelligence (AI), machine learning (ML) and allied technological advancements, especially in the mobile services space and setting the agenda for mobile disruption in India for 2019. Leading Indian startup leaders like Dr Pravin Bhagwat from Mojo Networks, Dr Manish Gupta, Founder of VideoKen, and Sourabh Issar, CEO of CloudSE were there to share their insights, along with corporate leaders like Dr Gautham Shroff of TCS Research, Jan Holler of Ericsson Research, Bhairav Acharya from Facebook, Shubhashis Sengupta from Accenture Labs and Vinutha BM from Wipro, as well as several students and researchers from the country's premier technological institutions.

China, the world's biggest agricultural producer, is leading the race when it comes to empowering farmers with Artificial Intelligence (AI)-driven technologies. The aim is clear: To help the community digitally record information to cut costs and increase yields -- with just a smartphone in their hands as AI leveraged Cloud computing to make sense of the data for farmers. India has now embarked on a journey to bring AI sensors into the fields. For Anant Maheshwari, the company's India President, Microsoft has begun empowering small-holder farmers in India to increase their income through higher crop yield and greater price control. "We are working with farmers, state governments, the Ministry of Electronics and Information Technology (MeitY) and the Ministry of Agriculture and Farmers Welfare to create an ecosystem for AI into farming," Maheshwari told IANS.

The People Matters TechHR Singapore 2019 will be a window to how today's incredible pace of technology advancement provides businesses, talent managers and technologists the perfect opportunity to engineer a fascinating workplace of tomorrow. One of the themes of the conference is harnessing technologies that could enable disproportionate leaps in work efficiency and nurturing a future ready digital workforce. One of the highlights of the event will be the keynote session spearheaded by top global tech influencer, Ray Wang. Ray Wang is the Principal Analyst, Founder, and Chairman of Silicon Valley-based Constellation Research, Inc. Wang has held executive roles in product, marketing, strategy, and consulting at companies such as Forrester Research, Oracle, PeopleSoft, Deloitte, Ernst & Young, and Johns Hopkins Hospital. Ray's expertise lies in digital, innovation, business model design, engagement strategies, customer experience, matrix commerce, and big data.

It is relatively well known that China has developed technology that can effectively monitor the country's citizens. Scientists working for The Chinese People's Liberation Army (PLA) have been working on CCTV cameras which have artificial intelligence embedded in them for years. The software, known as Ensemble Net, has been trained using clips from CCTV cameras to identify people and track their movements across the country. It's been tested across the country and, according to reports, it is supposedly 90 per cent accurate. The system will be able to scan body shapes and facial features, before scanning its database to see if it finds a match.

The goal of reinforcement learning is to construct algorithms that learn and plan in sequential decision making systems when the underlying system dynamics are unknown. A typical model in RL is Markov Decision Process (MDP). At each time step, the environment is in state s. The agent may take an action a, obtain a reward, and then the environment may transit to another state. In reinforcement learning, the transition probability distribution is unknown. The algorithm needs to learn the transition dynamics of MDP, while aiming to maximize the cumulative reward. This causes an exploration-exploitation dilemma: whether to act to gain new information (explore) or to act consistently with past experience to maximize reward (exploit). Theoretical analysis of reinforcement learning falls into two broad categories: those assuming a simulator (a.k.a.