Charge prediction, determining charges for criminal cases by analyzing the textual fact descriptions, is a promising technology in legal assistant systems. In practice, the fact descriptions could exhibit a significant intra-class variation due to factors like nonnormative use of language, which makes the prediction task very challenging, especially for charge classes with too few samples to cover the expression variation. In this work, we explore to use the charge definitions from criminal law to alleviate this issue. The key idea is that the expressions in a fact description should have corresponding formal terms in charge definitions, and those terms are shared across classes and could account for the diversity in the fact descriptions. Thus, we propose to create auxiliary fact representations from charge definitions to augment fact descriptions representation. The generated auxiliary representations are created through the interaction of fact description with the relevant charge definitions and terms in those definitions by integrated sentence-and word-level attention scheme. Experimental results on two datasets show that our model achieves significant improvement than baselines, especially for classes with few samples. Introduction The task of charge prediction is to determine appropriate charges, such as theft, seizing or robbery, for criminal cases by analyzing the textual fact descriptions.

Task oriented language understanding in dialog systems is often modeled using intents (task of a query) and slots (parameters for that task). Intent detection and slot tagging are, in turn, modeled using sentence classification and word tagging techniques respectively. Similar to adversarial attack problems with computer vision models discussed in existing literature, these intent-slot tagging models are often over-sensitive to small variations in input -- predicting different and often incorrect labels when small changes are made to a query, thus reducing their accuracy and reliability. However, evaluating a model's robustness to these changes is harder for language since words are discrete and an automated change (e.g. adding `noise') to a query sometimes changes the meaning and thus labels of a query. In this paper, we first describe how to create an adversarial test set to measure the robustness of these models. Furthermore, we introduce and adapt adversarial training methods as well as data augmentation using back-translation to mitigate these issues. Our experiments show that both techniques improve the robustness of the system substantially and can be combined to yield the best results.

While machine-type communication (MTC) devices generate considerable amounts of data, they often cannot process the data due to limited energy and computation power. To empower MTC with intelligence, edge machine learning has been proposed. However, power allocation in this paradigm requires maximizing the learning performance instead of the communication throughput, for which the celebrated water-filling and max-min fairness algorithms become inefficient. To this end, this paper proposes learning centric power allocation (LCPA), which provides a new perspective to radio resource allocation in learning driven scenarios. By employing an empirical classification error model that is supported by learning theory, the LCPA is formulated as a nonconvex nonsmooth optimization problem, and is solved by majorization minimization (MM) framework. To get deeper insights into LCPA, asymptotic analysis shows that the transmit powers are inversely proportional to the channel gain, and scale exponentially with the learning parameters. This is in contrast to traditional power allocations where quality of wireless channels is the only consideration. Last but not least, to enable LCPA in large-scale settings, two optimization algorithms, termed mirror-prox LCPA and accelerated LCPA, are further proposed. Extensive numerical results demonstrate that the proposed LCPA algorithms outperform traditional power allocation algorithms, and the large-scale algorithms reduce the computation time by orders of magnitude compared with MM-based LCPA but still achieve competing learning performance.

Project scheduling includes various problems of high pract ical relevance. Such problems arise in many areas and include different constraints and objectives. Usually pro ject scheduling problems require scheduling of a set of proj ect activities over a period of time and assignment of resources to these activities. Typical constraints include time windows for activities, precedence constraints between the ac tivities, assignment of appropriate resources etc. The aim is to find feasible schedules that optimize several criteria su ch as the minimization of total completion time. In this paper we investigate solving a real-world project sc heduling problem that arises in an industrial test laborato ry of a large company. This problem, Industrial Test Laborator y Scheduling (TLSP), which is an extension of the well known Resource-Constrained Project Scheduling Problem (R CPSP), was originally described in [1, 2]. It consists of a grouping stage, where smaller activities (tasks) are join ed into larger jobs, and a scheduling stage, where those jobs are scheduled and have resources assigned to them. In this wo rk, we deal with the second stage and assume that a grouping of tasks into jobs is already provided.

That's how much data humanity generates every single day. And the amount is increasing; we've created 90% of the world's data in the last two years alone. It should come as no surprise, then, that businesses today are drowning in data. That's because much of that data is unstructured; it takes the form of documents, social media content and other qualitative information that doesn't reside in conventional databases and is can't be parsed by traditional algorithms or machine analysis. Cognitive services not only cut through the deluge of data, but also bring meaning to it through human-like understanding of natural language queries.

His work is focused on the ethics of technology, including such topics as artificial intelligence (AI) and ethics, the ethics of space exploration and use, the ethics of technological manipulation of humans, the ethics of mitigation of and adaptation towards risky emerging technologies, and various aspects of the impact of technology and engineering on human life and society, including the relationship of technology and religion, particularly the Catholic Church. He spoke to Charles Camosy.] Camosy: Can you tell us how you became interested and indeed expert in AI ethics? Green: My undergraduate degree was in genetics from the University of California, Davis, and I worked in molecular biology and biotech there, but ultimately I discovered that lab work was not for me. I had dreamed of being a scientist since I was a child, so this was very confusing and I didn't know what to do with myself, so I made what turned out to be one of the best decisions of my life and joined the Jesuit Volunteers International.

The Artificial Intelligence (AI) in accounting market size is expected to grow from USD 666 million in 2019 to USD 4,791 million by 2024, at a Compound Annual Growth Rate (CAGR) of 48.4% during the forecast period. The AI in accounting is driven by various factors, such as the growing need to automate accounting processes and support enhanced data-based advisory and decision making. However, growing concerns over high criticality of data volume and quality, and investment related issues with integration of AI in accounting can hinder the growth of the market. Services segment to grow at a higher CAGR during the forecast period The AI in accounting market based on component is segmented into solutions and services.The services segment is expected to grow at a rapid pace during the forecast period. The growth of this segment can be attributed to the increasing deployment of AI in accounting software tools and solutions, which leads to increasing the demand for pre- and post-deployment services.

Are robots coming for your job? According to a Dell Technologies survey, 82% of leaders expect their employees and machines to work as "integrated teams". And many employees look forward to artificial intelligence (AI) that can help them do their job better. But not everybody has such a rosy outlook. In Australia, the report "Australia's Future Workforce" predicts about 40% of jobs could be lost to robotics, automation and artificial intelligence in the next 10-15 years.

Agricultural fields are no less than a battlefield. Irrespective of terrain, geography and type, crops have to compete against scores of different weeds, species of hungry insects, nematodes and a broad array of diseases. Weeds, or invasive plants, aggressively compete for soil nutrients, light and water, posing a serious threat to agricultural production and biodiversity. Weeds directly and indirectly result in tremendous losses to the farm sector, which convert to billions each year worldwide. To combat these challenges, the farm sector is looking at Artificial Intelligence (AI) based solutions.

Adobe and Evergage have topped the latest Forrester Wave rating digital intelligence platform providers based on product offering, strategy and market presence. The Q4 Digital Intelligence Platform report placed both Adobe and Evergage in the leaders' quadrant out a total field of nine vendors. In its commentary, the analyst firm highlighted Adobe's best-of-breed capabilities and scale as key reasons for its leadership position, noting the core components of Adobe Analytics, Experience Platform and Adobe Target as providing a wide array of services for marketing and customer engagement. Specifically, these address data, analytics and experience optimisation and are also well integrated into a wider platform play, Forrester stated. Second in the pack was smaller player, Evergage, which Forrester praised for its all-in-one, tightly integrated offering.