Due to the success of the pre-trained language model (PLM), existing PLM-based summarization models show their powerful generative capability. However, these models are trained on general-purpose summarization datasets, leading to generated summaries failing to satisfy the needs of different readers. To generate summaries with topics, many efforts have been made on topic-focused summarization. However, these works generate a summary only guided by a prompt comprising topic words. Despite their success, these methods still ignore the disturbance of sentences with non-relevant topics and only conduct cross-interaction between tokens by attention module. To address this issue, we propose a topic-arc recognition objective and topic-selective graph network. First, the topic-arc recognition objective is used to model training, which endows the capability to discriminate topics for the model. Moreover, the topic-selective graph network can conduct topic-guided cross-interaction on sentences based on the results of topic-arc recognition. In the experiments, we conduct extensive evaluations on NEWTS and COVIDET datasets. Results show that our methods achieve state-of-the-art performance.

Be prepared in the near future when you gaze into the blue skies to perceive a whole series of strange-looking things – no, they will not be birds, nor planes, or even superman. They may be temporarily, and in some cases startlingly mistaken as UFOs, given their bizarre and ominous appearance. But, in due course, they will become recognized as valuable objects of a new era of human-made flying machines, intended to serve a broad range of missions and objectives. Many such applications are already incorporated and well entrenched in serving essential functions for extending capabilities in our vital infrastructures such as transportation, utilities, the electric grid, agriculture, emergency services, and many others. Rapidly advancing technologies have made possible the dramatic capabilities of unmanned aerial vehicles (UAV/drones) to uniquely perform various functions that were inconceivable a mere few years ago.

IBM Data and AI general manager Rob Thomas discusses AI being incorporated into the workforce. The labor market may be humming right now, but there may be a dark cloud looming ahead. Over the course of the next decade, up to 800 million jobs globally could disappear due to advances in artificial intelligence and robotics, according to research from the McKinsey Global Institute, a top consulting firm. An estimated one-third of the 2030 workforce in the U.S. may need to learn new skills and find work in new occupations. The changes won't hit the country equally.

The use of robots in the workplace has more than double in just a 12 year period, displacing 50 percent of many human workers across the US, studies have found. A new interactive map provides more detail into this'robot exposure' by highlighting the top 10 metropolitan areas threatened by this machine takeover – California being listed as number one. In addition to areas most at risk, experts found that automation is displacing younger, less-educated and minority workers at the highest rates. The study and map were developed by The Century Foundation, a progressive think tank headquartered in New York City, which looked across more than 250 metropolitan areas to understand this'robot intensity'. Los Angeles, Long Beach and Santa Ana, California were ranked number one, followed by Chicago, Naperville and Joliet in Illinois.

With the advent of deep neural networks, some research focuses towards understanding their black-box behavior. In this paper, we propose a new type of self-interpretable models, that are, architectures designed to provide explanations along with their predictions. Our method proceeds in two stages and is trained end-to-end: first, our model builds a low-dimensional binary representation of any input where each feature denotes the presence or absence of concepts. Then, it computes a prediction only based on this binary representation through a simple linear model. This allows an easy interpretation of the model's output in terms of presence of particular concepts in the input. The originality of our approach lies in the fact that concepts are automatically discovered at training time, without the need for additional supervision. Concepts correspond to a set of patterns, built on local low-level features (e.g a part of an image, a word in a sentence), easily identifiable from the other concepts. We experimentally demonstrate the relevance of our approach using classification tasks on two types of data, text and image, by showing its predictive performance and interpretability.