Learning an optimal policy from a multi-modal reward function is a challenging problem in reinforcement learning (RL). Hierarchical RL (HRL) tackles this problem by learning a hierarchical policy, where multiple option policies are in charge of different strategies corresponding to modes of a reward function and a gating policy selects the best option for a given context. Although HRL has been demonstrated to be promising, current state-of-the-art methods cannot still perform well in complex real-world problems due to the difficulty of identifying modes of the reward function. In this paper, we propose a novel method called hierarchical policy search via return-weighted density estimation (HPSDE), which can efficiently identify the modes through density estimation with return-weighted importance sampling. Our proposed method finds option policies corresponding to the modes of the return function and automatically determines the number and the location of option policies, which significantly reduces the burden of hyper-parameters tuning. Through experiments, we demonstrate that the proposed HPSDE successfully learns option policies corresponding to modes of the return function and that it can be successfully applied to a challenging motion planning problem of a redundant robotic manipulator.

Large data collections required for the training of neural networks often contain sensitive information such as the medical histories of patients, and the privacy of the training data must be preserved. In this paper, we introduce a dropout technique that provides an elegant Bayesian interpretation to dropout, and show that the intrinsic noise added, with the primary goal of regularization, can be exploited to obtain a degree of differential privacy. The iterative nature of training neural networks presents a challenge for privacy-preserving estimation since multiple iterations increase the amount of noise added. We overcome this by using a relaxed notion of differential privacy, called concentrated differential privacy, which provides tighter estimates on the overall privacy loss. We demonstrate the accuracy of our privacy-preserving dropout algorithm on benchmark datasets.

Entertainment and advertising already pervade our homes and our smartphones. Before long, they'll be everywhere in our cars -- not just on the sound system and on little screens, but throughout the entire passenger compartment, even on the windows. Brian Krzanich, chief executive at computer chip maker Intel, on Wednesday announced a collaboration with Warner Bros. to create "immersive experiences" inside driverless cars. Speaking to auto industry insiders at Automobility LA -- the four-day preview event ahead of the Los Angeles Auto Show -- Krzanich said the companies will build proof-of-concept entertainment and advertising platforms using trademarked fictional characters to demonstrate how people might occupy themselves while a robot does the driving. Someone who otherwise would have been driving might instead pretend to be Batman, Krzanich said, as an augmented reality system projected images on windows to make it seem like the car was zipping through Gotham City. In a press release, Intel said those same windows "will enable passengers to view advertising and other discovery experiences."

Toyota's third-generation humanoid T-HR3 robot is remotely controlled by a an employee at the International Robot Exhibition in Tokyo on Nov. 29, 2017. Toyota Motor Corporation unveiled its third generation humanoid robot T-HR3, which is to be maneuvered seamlessly and in real time based on inputs from its human pilot, amongst other service and industrial robots at the International Robot Exhibition opening in Tokyo Wednesday.

Once autonomous cars can routinely take over the monotonous, minute-to-minute responsibilities of actually driving, what are their human passengers supposed to do with themselves? Well, if Intel has any say, people will spend their trips being immersed in interactive fantasy lands. To that end, Intel announced on Wednesday at the 2017 LA Auto Show that it will partner with Warner Bros. to develop "in-cabin, immersive experiences in autonomous vehicle (AV) settings," according to the company's press release. Essentially, not only will passengers be able to watch movies, TV shows or play games on their mobile devices while their autonomous vehicles are driving, they'll eventually engage with fully immersive VR and AR experiences as well. "For example, a fan of the superhero Batman could enjoy riding in the Batmobile through the streets of Gotham City, while AR capabilities render the car a literal lens to the outside world," Intel CEO Brian Krzanich wrote, "enabling passengers to view advertising and other discovery experiences."

As our world becomes more and more technology-driven, robots could replace workers in a huge number of jobs, a new report has warned. The report claims that as many as 800 million workers could be replaced by machines in just 13 years. Jobs most likely to be taken include fast-food workers and machine-operators, while gardeners, plumbers and childcare workers are the least likely to be replaced by bots, according to the report. In terms of jobs, the report suggests that physical jobs in predictable environments – including machine-operators and fast-food worker – are the most likely to be replaced by robots. But it added: 'Collecting and processing data are two other categories of activities that increasingly can be done better and faster with machines.

Last winter, the sportswear giant Adidas opened a pop-up store inside a Berlin shopping mall. The boutique was part of a corporate experiment called Storefactory--a name as flatly self- explanatory as it is consistent with the convention of German compound nouns. It offered a single product: machine- knit merino wool sweaters, made to order on the spot. Customers stepped up for body scans inside the showroom and then worked with an employee to design their own bespoke pullovers. The sweaters, which cost the equivalent of about $250 apiece, then materialized behind a glass wall in a matter of hours.

The world is commonly divided into industrialized and emerging economies. A new study of how technology will transform demand for workers suggests we might talk of the automated and automating worlds instead. Economic think tank McKinsey Global Institute forecast changes in demand for different kinds of labor across 45 countries as technologies improve to perform physical or office tasks. One key result: Robots pose a more immediate and disruptive threat to the US middle class than they do to middle-income workers in less developed countries like India. The report warns that in the US technology will crimp demand for many types of work, such as office administration and operating construction equipment.

To find the best innovations in retail banking, you usually need to look beyond North America. The best evidence of this ongoing trend is a review of winners in major financial innovation competitions worldwide. Here is a summary of some of the best-of-the-best innovations recognized by Efma and Accenture. The question that gets asked at almost every gathering of financial services executives is, "What institution is the best innovator in banking?" or "Where is most innovation in banking taking place?" Luckily there are trade organizations such as Efma (an association of 3,300 financial institutions in 130 countries) and the Bank Administration Institute (BAI) that have annual competitions to recognize the best in the financial services industry.

The government is continuing the process of data mining of deregistered companies and so far, bank details have been gathered for nearly 50,000 such entities, Union minister P P Chaudhary said today. Amid the clampdown on the black money menace, names of more than 2.24 lakh companies have been struck off from the records and over 3 lakh directors have been barred from directorship for their associations with such firms. The minister of state for corporate affairs said that based on details gathered from banks, around 50,000 deregistered companies deposited and withdrew about Rs 17,000 crore during demonetisation. Data mining is continuing with respect to the struck-off entities, Chaudhary said, adding that artificial intelligence could be used to identify illegal activities of companies. He was speaking at an event organised by the Institute of Cost Accountants of India.