Methods based on vector embeddings of knowledge graphs have been actively pursued as a promising approach to knowledge graph completion.However, embedding models generate storage-inefficient representations, particularly when the number of entities and relations, and the dimensionality of the real-valued embedding vectors are large. We present a binarized CANDECOMP/PARAFAC(CP) decomposition algorithm, which we refer to as B-CP, where real-valued parameters are replaced by binary values to reduce model size. Moreover, we show that a fast score computation technique can be developed with bitwise operations. We prove that B-CP is fully expressive by deriving a bound on the size of its embeddings. Experimental results on several benchmark datasets demonstrate that the proposed method successfully reduces model size by more than an order of magnitude while maintaining task performance at the same level as the real-valued CP model.

Achieving fusion of deep learning with combinatorial algorithms promises transformative changes to artificial intelligence. One possible approach is to introduce combinatorial building blocks into neural networks. Such end-to-end architectures have the potential to tackle combinatorial problems on raw input data such as ensuring global consistency in multi-object tracking or route planning on maps in robotics. In this work, we present a method that implements an efficient backward pass through blackbox implementations of combinatorial solvers with linear objective functions. We provide both theoretical and experimental backing. In particular, we incorporate the Gurobi MIP solver, Blossom V algorithm, and Dijkstra's algorithm into architectures that extract suitable features from raw inputs for the traveling salesman problem, the min-cost perfect matching problem and the shortest path problem.

Machine learning techniques are currently used extensively for automating various cybersecurity tasks. Most of these techniques utilize supervised learning algorithms that rely on training the algorithm to classify incoming data into different categories, using data encountered in the relevant domain. A critical vulnerability of these algorithms is that they are susceptible to adversarial attacks where a malicious entity called an adversary deliberately alters the training data to misguide the learning algorithm into making classification errors. Adversarial attacks could render the learning algorithm unsuitable to use and leave critical systems vulnerable to cybersecurity attacks. Our paper provides a detailed survey of the state-of-the-art techniques that are used to make a machine learning algorithm robust against adversarial attacks using the computational framework of game theory. We also discuss open problems and challenges and possible directions for further research that would make deep machine learning-based systems more robust and reliable for cybersecurity tasks.

New South Wales in Australia has rolled out high-definition cameras to catch people using their cellphones while driving, according to media reports. The technology is intended to target illegal use of cellphones through fixed and mobile trailer-mounted cameras, New South Wales Minister for Roads Andrew Constance was quoted as saying by the CNN. The cameras use artificial intelligence to scan images and zero in on the offenders. The identified images will be verified by authorised personnel, and the images will be securely stored and managed, authorities said. As many as 45 portable cameras will be set up across the Australian state at unknown locations and without warning signs in the next three years, CNN affiliate Sky News Australia reported.

Artificial intelligence is intended to mimic the behavior of humans. However, AI technology can be applied in reverse – it can be used to change human behavior. There are a number of reasons AI can be great for social engineering, especially in the field of marketing. Julita Vassileva, a Professor at the University of Saskatchewan in Canada wrote a great paper on AI for Human Learning and Behavior. This paper was published in Frontiers in Artificial Intelligence Journal.

The fast-changing, uncertain and ambiguous environments that organisations operate in today, require organisations to re-think all their internal business processes and customer touchpoints. Besides, due to the availability of Emerging Information Technologies such as big data analytics, blockchain and artificial intelligence, it has become easier for startups to compete with existing organisations. Often these startups are more flexible and agile than Fortune 1000 companies, and they can become a significant threat if not paid attention to. Therefore, focusing purely on the day-to-day operation is simply not enough and organisations have to become innovative and adaptive to change if they wish to remain competitive. When big data, blockchain and AI are combined, it will change collaboration among individuals, organisations and things, moving from pure human-to-human collaboration to increasingly human-to-machine collaboration and machine-to-machine collaboration. It will turn every organisation into a data organisation.

Princeton, New Jersey--(Newsfile Corp. - December 3, 2019) - NotForgotten Digital Library LLC, creators of the first digital time capsule powered by blockchain, today announced their use of Veritone aiWARE and Digital Media Hub. Veritone's cognitive capabilities will add a new dimension to NotForgotten's current offerings by adding automated video transcription and metadata creation to make large volumes of video content easily searchable. Videos within NotForgotten archives can then be easily accessed through the custom-branded Digital Media Hub and users can discover insights about a set time and place in history. "The new capabilities Veritone is adding to NotForgotten's archives are set to change the way future historians study our present," says Adrienne Liebenberg, co-founder at NotForgotten. "Veritone aiWARE will analyze the videos within the NotForgotten Digital Library and add AI-enabled insights about society and people within a set time and place. This will be the first time we - society at large - will have the framework in place to not only record and research history from a personal perspective, but to also see how these events made people feel."

A new leak of highly classified Chinese government documents has uncovered the operations manual for running the mass detention camps in Xinjiang and exposed the mechanics of the region's Orwellian system of mass surveillance and "predictive policing." The China Cables, obtained by the International Consortium of Investigative Journalists, include a classified list of guidelines, personally approved by the region's top security chief, that effectively serves as a manual for operating the camps now holding hundreds of thousands of Muslim Uighurs and other minorities. The leak also features previously undisclosed intelligence briefings that reveal, in the government's own words, how Chinese police are guided by a massive data collection and analysis system that uses artificial intelligence to select entire categories of Xinjiang residents for detention. The manual, called a "telegram," instructs camp personnel on such matters as how to prevent escapes, how to maintain total secrecy about the camps' existence, methods of forced indoctrination, how to control disease outbreaks, and when to let detainees see relatives or even use the toilet. The document, dating to 2017, lays bare a behavior-modification "points" system to mete out punishments and rewards to inmates. The manual reveals the minimum duration of detention: one year -- though accounts from ex-detainees suggest that some are released sooner. Experts say the platform, which is used in both policing and military contexts, demonstrates the power of technology to help drive industrial-scale human rights abuses. The China Cables reveal how the system is able to amass vast amounts of intimate personal data through warrantless manual searches, facial recognition cameras, and other means to identify candidates for detention, flagging for investigation hundreds of thousands merely for using certain popular mobile phone apps.

In 2010, US authors in top-rated AI journals outnumbered Chinese counterparts by two to one. That ratio has now reversed. Last year, 1,073 AI experts based at Chinese universities were credited in AI journals such as the Institute of Electrical and Electronics Engineers's Transactions on Neural Networks, compared to 492 US authors. Australia and Israel also do well on this metric. When experts are ranked according to their'H-index' – a metric of productivity and the citation impact of the publications of a scientist or scholar – Americans occupy 626 of the 1,000 top spots, including all of the top ten spots at the time of our analysis. New Zealand, Saudi Arabia and Finland's AI academics are also highly ranked.

Public-sector CIOs can use this list of strategic technology trends as inspiration and factor them into their organizations' strategic plans. Forward-looking government officials know that, in a digital society, "Policy is the technology and technology is the policy." Any government service delivered at scale is underpinned by a host of technologies. If the success of these business projects is compromised by poor implementation of technology, then the political objectives are compromised too. Implementing a digital government strategy is a journey that will span multiple budget cycles and political administrations.