Extractive summarization models require sentence-level labels, which are usually created heuristically (e.g., with rule-based methods) given that most summarization datasets only have document-summary pairs. Since these labels might be suboptimal, we propose a latent variable extractive model where sentences are viewed as latent variables and sentences with activated variables are used to infer gold summaries. During training the loss comes \emph{directly} from gold summaries. Experiments on the CNN/Dailymail dataset show that our model improves over a strong extractive baseline trained on heuristically approximated labels and also performs competitively to several recent models.

This letter is concerned with a recently developed paradigm of population-based optimization, termed particle filter optimization (PFO). In contrast with the commonly used meta-heuristics based methods, the PFO paradigm is attractive in terms of coherence in theory and easiness in mathematical analysis and interpretation. However, current PFO algorithms only work for single-objective optimization cases, while many real-life problems involve multiple objectives to be optimized simultaneously. To this end, we make an effort to extend the scope of application of the PFO paradigm to multi-objective optimization (MOO) cases. An idea called path sampling is adopted within the PFO scheme to balance the different objectives to be optimized. The resulting algorithm is thus termed PFO with Path Sampling (PFOPS). Experimental results show that the proposed algorithm works consistently well for three different types of MOO problems, which are characterized by an associated convex, concave and discontinuous Pareto front, respectively.

Character-level features are currently used in different neural network-based natural language processing algorithms. However, little is known about the character-level patterns those models learn. Moreover, models are often compared only quantitatively while a qualitative analysis is missing. In this paper, we investigate which character-level patterns neural networks learn and if those patterns coincide with manually-defined word segmentations and annotations. To that end, we extend the contextual decomposition technique (Murdoch et al. 2018) to convolutional neural networks which allows us to compare convolutional neural networks and bidirectional long short-term memory networks. We evaluate and compare these models for the task of morphological tagging on three morphologically different languages and show that these models implicitly discover understandable linguistic rules. Our implementation can be found at https://github.com/FredericGodin/ContextualDecomposition-NLP .

A challenge in creating a dataset for machine reading comprehension (MRC) is to collect questions that require a sophisticated understanding of language to answer beyond using superficial cues. In this work, we investigate what makes questions easier across recent 12 MRC datasets with three question styles (answer extraction, description, and multiple choice). We propose to employ simple heuristics to split each dataset into easy and hard subsets and examine the performance of two baseline models for each of the subsets. We then manually annotate questions sampled from each subset with both validity and requisite reasoning skills to investigate which skills explain the difference between easy and hard questions. From this study, we observed that (i) the baseline performances for the hard subsets remarkably degrade compared to those of entire datasets, (ii) hard questions require knowledge inference and multiple-sentence reasoning in comparison with easy questions, and (iii) multiple-choice questions tend to require a broader range of reasoning skills than answer extraction and description questions. These results suggest that one might overestimate recent advances in MRC.

Experts are developing a robot to replace human strawberry pickers as farms struggle to find workers due to Brexit. Around 20 per cent of soft fruits are going to waste due to a shortage of workers, University of Essex researchers say. This will worsen when Britain leaves the EU, scientist claim, which has led to farms looking for alternate solutions to harvest crops. Dr Vishuu Mohan a computer science and engineering lecturer who is leading the project, said: 'The challenge is that no two berries are the same - they come in different shapes, sizes, order of ripeness and many are hidden in the foliage. 'Also the environment keeps changing constantly - sunny, windy, rainy - in contrast to a typical industrial environment.

Several studies have shown that by the end of 2030, more than 30% of the current jobs will be unnecessary – due to technological advancements. AI and machine learning will, undoubtedly, change the way job descriptions are designed and will fundamentally modify the manner in which humans carry their work. PricewaterhouseCoopers recently published a study; according to the study, more 35% (approximately 38%, more exactly) of the jobs available on the US market could be endangered by automation and AI's immersion into our daily lives. The estimations are similar for the rest of the world, as well. More than 35% of the jobs available on the German market will also be replaced by machines and AI, while in the UK, the ratio is expected to revolve somewhere around 30%.

Amazon has opened a second location for its radical Amazon Go concept store. The store, the second in Amazon's hometown of Seattle, will be a mile away from the original location near the Seattle Central Library at 920 Fifth Ave., and opened at 7 a.m. The new store is slightly smaller than the original, at 1,450 square feet, and won't sell alcohol or staples like milk and bread. Amazon confirmed the new store in a statement, saying'We are excited to bring Amazon Go to 920 5th Avenue in Seattle. The store will open in Fall 2018.' Pictured, the original store It also forgoes an in store kitchen, and will have its fresh food supplied by an Amazon kitchen facility in Seattle. Gianna Puerini, the Amazon vice president who oversees Go, told the Seattle Times she expects a higher portion of office workers among the clientele compared to the first store.

DeepMind Technologies, an artificial intelligence (AI) company based in London (U.K.) that was purchased by Google in 2014, recently announced that it is applying its machine learning algorithms to Google data centers to reduce the amount of energy dedicated to cooling by as much as 40 percent.

Workshop 18 is designated by Chinese industry officials as a model demonstration facility for Beijing's plan to upgrade its corporate champions so they can better compete in the world, a policy known as "Made in China 2025." The adoption of robots, big data and other technological advancements is seen by Chinese leaders as key to developing domestic giants in areas such as power equipment, electric cars, marine products and chips--sectors it aims to become largely self-sufficient by the middle of the next decade. Sany, one of China's three big heavy machinery makers, said the integration of technology has increased capacity, shortened order-delivery times and slashed operational costs, all by at least 20%. The company is also betting that the technology will enable it to build a reputation for innovation and quality, rather than for lower prices for copycat products that made Sany a big player domestically. "The future of the heavy-equipment industry will rely as much on software and data as it does on hardware," Sany's Chief Information Officer Pan Ruigang said. The "Made in China 2025" strategy has drawn criticism in Washington, with members of President Trump's administration accusing Beijing of using subsidies and protectionism to unfairly bolster Chinese companies.

It's been more than a century since Spanish neuroanatomist Santiago Ramón y Cajal won the Nobel Prize for illustrating the way neurons allow you to walk, talk, think, and be. In the intervening hundred years, modern neuroscience hasn't progressed that much in how it distinguishes one kind of neuron from another. Sure, the microscopes are better, but brain cells are still primarily defined by two labor-intensive characteristics: how they look and how they fire. Which is why neuroscientists around the world are rushing to adopt new, more nuanced ways to characterize neurons. Sequencing technologies, for one, can reveal how cells with the same exact DNA turn their genes on or off in unique ways--and these methods are beginning to reveal that the brain is a more diverse forest of bristling nodes and branching energies than even Ramón y Cajal could have imagined.