Some people worry that hackers could infiltrate their smart speakers and spy on them, but that hasn't been the practical reality -- not for Amazon's Echo, at least. A team of researchers from China's Tencent has come about as close as you can get right now, however. They've disclosed an attack on the Echo that uses both a modified speaker and a string of Alexa web interface vulnerabilities to remotely eavesdrop on regular models. It sounds nefarious, but it requires more steps than would be viable for most intruders. The team created a rogue Echo by removing a flash memory chip from the device, modifying its firmware to get root access, and soldering it back on its circuit board.

For many organisations and their workers, technology automation is an opportunity and a threat. On one hand, using robotic process automation and other tools creates efficiencies by automating manual and repetitive tasks. On the other, it has the potential to make jobs redundant. Tech execs gathered at roundtable events in Sydney and Melbourne recently to discuss the impact automation technologies are having on their organisations. A key issue raised during the discussion was the challenge of ensuring governance policies and procedures are followed as'bots' replace tasks carried out by humans across the business.

Konica Minolta Japan is the distributor and service provider arm of Konica Minolta Inc., which primarily manufactures multifunctional peripherals, like copiers and digital print systems. The company's goal is to develop a one-stop organization where customers can solve their office and printing issues, thereby fostering trust and loyalty. An industry leader, Konica Minolta Japan believes in the importance of incorporating AI and sensor data into its operations. The company implemented SAS to enhance business efficiency and customer satisfaction. Shouichi Yabe, Director of the Data Science Implementation Group, oversaw Konica Minolta Japan's SAS AI and IoT project.

Security researchers at Trustwave have released a new open-source tool that uses facial recognition technology to locate targets across numerous social media networks on a large scale. Dubbed Social Mapper, the facial recognition tool automatically searches for targets across eight social media platforms, including--Facebook, Instagram, Twitter, LinkedIn, Google, the Russian social networking site VKontakte, and China's Weibo and Douban--based on their names and pictures. The tool's creators claim they developed Social Mapper intelligence-gathering tool predominantly to help pen testers and red teamers with social engineering attacks. Although the searches of names and pictures can already be performed manually, Social Mapper makes it possible to automate such scans far faster and "on a mass scale with hundreds or thousands of individuals" at once. "Performing intelligence gathering online is a time-consuming process, it typically starts by attempting to find a person's online presence on a variety of social media sites," Trustwave explained in a blog post detailing the tool.

As artificial intelligence continues to evolve, it is having profound impact on a range of sectors seemingly unrelated to it, such as international relations. Some countries are pursuing AI more or less within the confines of international law and generally accepted principles of doing business, while others are choosing to do what is necessary to attempt to achieve AI supremacy outside those boundaries. In the process, AI is slowly altering the balance of power between global actors and among alliances in a number of ways. Just as becoming adept in the cyber arena levels the playing field – giving countries such as Iran and North Korea the ability to go head to head with China, Russia and that US in cyber space – the pursuit of AI supremacy is providing an increased competitive edge in international business to some smaller, otherwise less competitive nations, enhancing their ability to secure preferential trade and investment arrangements with other countries, raising their global profile, and enabling them to progress into previously unimagined areas of international trade, investment, and diplomacy. How AI is deployed by governments can have serious consequences in international relations, particularly if a given government has unusual capabilities in the AI arena.

CELCOM Axiata Bhd inked an agreement with Huawei Technologies (Malaysia) Sdn Bhd to apply the Cloud-based Digitised Operation Platform, Software as a Service (SaaS) solution. Celcom will be the first in the country to adopt full suite Cloud-based Operation Support Service (OSS) system to accelerate agility in their automation and intelligence of network management, and pave the way for their journey towards becoming a digital company. The Digitised Operation Platform brings together Artificial Intelligence (AI) and Machine Learning technology powered by Huawei's Operation Web Services (OWS) suite, to enhance Celcom's capabilities in managing increasingly complex networks and services. It also enables Celcom to transform their daily operations from reactive to proactive and predictive, and further solidify their drive to deliver an awesome customer experience. The agreement to acquire the platform for Celcom's network operation was signed by Celcom Axiata chief technology officer Amandeep Singh and Huawei Technologies (Malaysia) chief executive officer Baker Zhouxin. Through this partnership, Huawei aims to leverage on its Digitised Operation AUTomation & INtelligence Services Solution (AUTIN), and share global experiences with Celcom to achieve a visualised, automated and intelligent network operation.

Summarizing, there are four desiderata that one would like to see in A.I. systems, if they have to compare well with human (or just animal) brains: meta-learning, learning by demonstration ("few-shot learning"), transfer learning and multi-task learning. Meta-learning is particularly relevant in the case of reinforcement learning. It is obvious that reinforcement learning is highly unnatural. DeepMind's AlphaGo and OpenAi Five need to learn from scratch via a huge number of trials. Animals, instead, use built-in or acquired "meta-skills" to learn new tasks in just a few trials. Modern computational theory of meta-learning (learning how to learn) dates back at least to the 1990s, when Schmidhuber published the manifesto "Simple Principles of Metalearning" (1996), followed by his student Sepp Hochreiter ("Learning to Learn Using Gradient Descent", 2001), and by Nicolas Schweighofer and Kenji Doya at Japan's ATR ("Meta-learning in Reinforcement Learning", 2001). Examples of "deep" meta-learning systems of the new generation are: RL Square by Pieter Abbeel's student Yan Duan at UC Berkeley, based on Schulman's TRPO ("RL Square: Fast Reinforcement Learning via Slow Reinforcement Learning", 2016); the "model-agnostic meta-learning" (MAML) of Sergey Levine's student Chelsea Finn at UC Berkeley ("Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks", 2017); Marcel Binz's thesis at KTH Royal Institute of Technology ("Learning Goal-Directed Behaviour", 2017); Jane Wang's "deep meta-reinforcement learning" at DeepMind ("Learning to Reinforcement Learn", 2017); and OpenAI's Reptile, developed by Alex Nichol and John Schulman, a generalization of Finn's MAML ("On First-Order Meta-Learning Algorithms", 2018). DeepMind's neuroscientist Matthew Botvinick believes that the latter could be a model for how our brain learns: the dopamine system trains another part of the brain, the prefrontal cortex, to operate as its own free-standing learning system ("Prefrontal Cortex as a Meta-reinforcement Learning System", 2018).

Background: It is still an open research area to theoretically understand why Deep Neural Networks (DNNs)-- equipped with many more parameters than training data and trained by (stochastic) gradient-based methods-- often achieve remarkably low generalization error. Contribution: We study DNN training by Fourier analysis. Our theoretical framework explains: i) DNN with (stochastic) gradient-based methods endows low-frequency components of the target function with a higher priority during the training; ii) Small initialization leads to good generalization ability of DNN while preserving the DNN's ability of fitting any function. These results are further confirmed by experiments of DNNs fitting the following datasets, i.e., natural images, one-dimensional functions and MNIST dataset.

Acute Kidney Injury (AKI), a sudden decline in kidney function, is associated with increased mortality, morbidity, length of stay, and hospital cost. Since AKI is sometimes preventable, there is great interest in prediction. Most existing studies consider all patients and therefore restrict to features available in the first hours of hospitalization. Here, the focus is instead on rehospitalized patients, a cohort in which rich longitudinal features from prior hospitalizations can be analyzed. Our objective is to provide a risk score directly at hospital re-entry. Gradient boosting, penalized logistic regression (with and without stability selection), and a recurrent neural network are trained on two years of adult inpatient EHR data (3,387 attributes for 34,505 patients who generated 90,013 training samples with 5,618 cases and 84,395 controls). Predictions are internally evaluated with 50 iterations of 5-fold grouped cross-validation with special emphasis on calibration, an analysis of which is performed at the patient as well as hospitalization level. Error is assessed with respect to diagnosis, race, age, gender, AKI identification method, and hospital utilization. In an additional experiment, the regularization penalty is severely increased to induce parsimony and interpretability. Predictors identified for rehospitalized patients are also reported with a special analysis of medications that might be modifiable risk factors. Insights from this study might be used to construct a predictive tool for AKI in rehospitalized patients. An accurate estimate of AKI risk at hospital entry might serve as a prior for an admitting provider or another predictive algorithm.

Yinlam Chow DeepMind We study risk-sensitive imitation learning where the agent's goal is to perform at least as well as the expert in terms of a risk profile. We first formulate our risk-sensitive imitation learning setting. We consider the generative adversarial approach to imitation learning (GAIL) and derive an optimization problem for our formulation, which we call risk-sensitive GAIL (RS-GAIL). We then derive two different versions of our RS-GAIL optimization problem that aim at matching the risk profiles of the agent and the expert w.r.t. Jensen-Shannon (JS) divergence and Wasserstein distance, and develop risk-sensitive generative adversarial imitation learning algorithms based on these optimization problems. We evaluate the performance of our JSbased algorithm and compare it with GAIL and the risk-averse imitation learning (RAIL) algorithm in two Mu-JoCo tasks.