Asia
Knowledge Aware Conversation Generation with Explainable Reasoning on Augmented Graphs
Liu, Zhibin, Niu, Zheng-Yu, Wu, Hua, Wang, Haifeng
Two types of knowledge, triples from knowledge graphs and texts from unstructured documents, have been studied for knowledge aware open-domain conversation generation, in which triple attributes or graph paths can narrow down vertex candidates for knowledge selection decision, and texts can provide rich information for response generation. Fusion of a knowledge graph and texts might yield mutually reinforcing advantages for conversation generation, but there is less study on that. To address this challenge, we propose a knowledge aware chatting machine with three components, an augmented knowledge graph containing both triples and texts, knowledge selector, and response generator. For knowledge selection on the graph, we formulate it as a problem of multi-hop graph reasoning that is more explainable and flexible in comparison with previous works. To fully leverage long text information that differentiates our graph from others, we improve a state of the art reasoning algorithm with machine reading comprehension technology. We demonstrate that supported by such unified knowledge and explainable knowledge selection method, our system can generate more appropriate and informative responses than baselines.
Improving Interactive Reinforcement Agent Planning with Human Demonstration
Li, Guangliang, Gomez, Randy, Nakamura, Keisuke, Lin, Jinying, Zhang, Qilei, He, Bo
TAMER has proven to be a powerful interactive reinforcement learning method for allowing ordinary people to teach and personalize autonomous agents' behavior by providing evaluative feedback. However, a TAMER agent planning with UCT---a Monte Carlo Tree Search strategy, can only update states along its path and might induce high learning cost especially for a physical robot. In this paper, we propose to drive the agent's exploration along the optimal path and reduce the learning cost by initializing the agent's reward function via inverse reinforcement learning from demonstration. We test our proposed method in the RL benchmark domain---Grid World---with different discounts on human reward. Our results show that learning from demonstration can allow a TAMER agent to learn a roughly optimal policy up to the deepest search and encourage the agent to explore along the optimal path. In addition, we find that learning from demonstration can improve the learning efficiency by reducing total feedback, the number of incorrect actions and increasing the ratio of correct actions to obtain an optimal policy, allowing a TAMER agent to converge faster.
The problem with AI? Study says it's too white and male, calls for more women, minorities
The ACLU and other groups urged Amazon to halt selling facial recognition technology to law enforcement departments. Lending tools charge higher interest rates to Hispanics and African Americans. Job hunting tools favor men. Negative emotions are more likely to be assigned to black men's faces than white men. Computer vision systems for self-driving cars have a harder time spotting pedestrians with darker skin tones.
Nidec to acquire Omron's auto electronics unit for ¥100 billion
OSAKA - Electric motor-maker Nidec Corp. said Tuesday it will acquire component-maker Omron Corp.'s automotive electronics subsidiary for ¥100 billion ($893 million), as it seeks to speed up the development of technology for autonomous and other advanced vehicles. Kyoto-headquartered Nidec will take an entire stake in Omron Automotive Electronics Co. by the end of October, as it aims to combine its strengths in motors, radar and camera-related technologies with Omron Automotive's edge in auto components for self-driving vehicles. "We want to widen our product lineup through the acquisition and enhance our competitiveness in the automobile-oriented business," as the auto industry has been shifting its focus to more electrified and self-driving vehicles, Nidec Chairman and CEO Shigenobu Nagamori told a news conference in Tokyo. Omron said it will focus more on its industrial automation and health care businesses as it found it difficult to continue hefty investments in developing auto technology on its own amid the intensifying race to make next-generation vehicles. Nidec, founded in 1973, has grown in part due to its ambitious mergers and acquisitions strategy.
PS5 release date: Sony hints at when new PlayStation will come out
The PlayStation 5 isn't coming this year but could arrive soon after that, Sony has suggested. The console is coming along quickly, with many of its specs in place and developers already working with early versions to understand the kind of games they might be able to create, its architect Mark Cerny has revealed in a wide-ranging interview with Wired. Mr Cerny explicitly said that the console won't be arriving in 2019, in line with expectations but dashing the hopes of anyone who had hoped the console could be about to arrive by surprise. We'll tell you what's true. You can form your own view.
Xbox One S All-Digital Edition: Microsoft reveals new, disc-less version of console
Microsoft has revealed a new version of the Xbox – missing what until now was one of its central features. The Xbox One S All-Digital Edition comes without a disc drive, meaning that people are only able to play games or films that they have downloaded over the internet, and are unable to use discs. In exchange for that missing feature, Microsoft says the new Xbox will always be at least $50 cheaper than the normal version of the Xbox One S. The All-Digital edition will cost $250 initially, but will receive price cuts whenever the traditional version does, the company indicated. We'll tell you what's true. You can form your own view.
Apple settles dispute with Qualcomm, potentially allowing new features to come to the iPhone
Apple has settled a major argument with chip maker Qualcomm that could help change the future of the iPhone. The two companies were locked in a bitter and worldwide international dispute about the technology that iPhones use to connect to the internet. The pair had been expected to try and resolve the dispute in legal hearings in San Diego, in a case that involved Apple's key iPhone suppliers. But just as that case began, the surprise truce was announced, with few details of the settlement being revealed. We'll tell you what's true.
UK porn ban to come into effect in July
The UK's porn ban will go into effect in July, the government has announced. After 15 July, all internet users will have be forced to prove themselves to be over 18 or be entirely blocked from seeing adult content. Under the new rules, any commercial provider of online pornography will have to carry out "robust age-verification checks on users" to ensure they are adults. We'll tell you what's true. You can form your own view.
UK porn ban condemned as 'dangerous' and 'irresponsible' amid scam warning
The UK's porn ban is "disastrous" for people's privacy, critics have warned. A ban on pornographic websites will come into effect on 15 July, meaning any visitor will need to prove they are over 18 through "robust age-verification checks on users". These could take the form of entering personal details online, or even buying a pass from local shops. We'll tell you what's true. You can form your own view.
What robots and AI may mean for university lecturers and students
The number of robots around the world is increasing rapidly. And it's said that automation will threatening more than 800m jobs worldwide by 2030. In the UK, it's claimed robots will replace 3.6m workers by this date, which means one in five British jobs would be performed by an intelligent machine. Jobs in higher education are no exception – with recent studies showing a rapid advancement in the use of these technologies in universities. The full potential of these disruptive technologies is yet to be discovered, but their impact on teaching and learning is expected to be huge.