PARIS – Coastal areas currently home to 300 million people will be vulnerable by 2050 to flooding made worse by climate change, no matter how aggressively humanity curbs carbon emissions, scientists said Tuesday. By midcentury and beyond, however, choices made today will determine whether Earth's coastlines remain recognizable to future generations, they reported in the journal Nature Communications. Destructive storm surges fueled by increasingly powerful cyclones and rising seas will hit Asia hardest, according to the study. More than two-thirds of the populations at risk are in China, Bangladesh, India, Vietnam, Indonesia and Thailand. Using a form of artificial intelligence known as neural networks, the new research corrects ground elevation data that have up to now vastly underestimated the extent to which coastal zones are subject to flooding during high tide or major storms. "Sea-level projections have not changed," co-author Ben Strauss, chief scientist and CEO of Climate Central, a U.S.-based non-profit research group, told AFP. "But when we use our new elevation data, we find far more people living in vulnerable areas that we previously understood."

This blog post was guest-written by Annie O'Rourke, CEO of Digital Workforce Australia and 89 Degrees East. She will be guest speaking about'Addressing real world problems with Artificial Intelligence' session of the Women Rock-IT series on 17 October. To sign up for this or other webinars in the series, click here. Don't tell my husband, but I've recently started an affair. No need to be too shocked though, because I'm pretty sure I can package it as so-called ethical polygamy.

Tinder, the most popular dating app in the world, has banned teens under the age of 18 but it's not stopping them from signing up. A jilted Australian doctor pleaded guilty Monday to sending 9,000 abusive and threatening messages to her former Tinder date, according to a new report. Radiologist Denise Jane Lee, 40, of Sydney, copped to four of 10 charges against her ahead of a scheduled five-day hearing in the Downing Centre Local Court, the Australian Associated Press reported. Lee, who was arrested in February 2017, copped to three counts of using a carriage service to harass, menace or offend and one count of intimidation, according to the report. Six additional charges were withdrawn.

According to an old fable, if you sat a monkey at a keyboard and gave it enough time, it would eventually type out the entire works of Shakespeare. The idea behind the thought experiment is a simple one, but it is an outdated way of thinking about artificial intelligence (AI). For example, consider the predictive text function on a phone -- unlike the hypothetical monkey, it does not randomly generate suggested words, but detects patterns in the way we write to select the likeliest options. "We've gone from predicting the next letters in a word to predicting the next word to being able to autogenerate whole sentences, and sometimes whole articles," Adelaide-based AI researcher, Dr John Flackett, said. Last year, Google launched Smart Compose for Gmail users which anticipates potential endings to sentences as you type them out, with the company promising it would "save you time".

The tech giants Amazon, Google and Facebook have all begun to use machine learning to give you tips on what to wear. Is fashion styling the next field to be disrupted by artificial intelligence (AI), or will the human eye remain supreme? It's too soon to know for sure, but understanding what machine learning is good at and how that overlaps with what fashion is all about can help us make some educated guesses. One thing machine learning does very well is finding patterns and common features among groups of items. Taking advantage of this, Google Lens and Amazon Style Snap can each identify a garment from a photo or video and then tell you a bit more, like how other people have worn it or where you can buy it.

Tuberculosis (TB) can now be identified in less than an hour thanks to a new blood test. The test procedure -- developed by The University of Queensland's Emeritus Professor Ian Riley in collaboration with researchers in Tanzania, India, Mexico and the Philippines -- is hoped to positively impact TB diagnosis in adults living in remote areas. "TB has been difficult to control because its symptoms are similar to many other diseases," Prof Riley said. "Other challenges include drug resistance to the disease and the high burden of HIV-positive cases in developing countries." Prof Riley explained that the discovery of the testing procedure came from using machine learning techniques to study three groups of adults who had a persistent cough for more than three weeks.

Autonomous Vehicles (AVs) are increasingly embraced around the world to advance smart mobility and more broadly, smart, and sustainable cities. Algorithms form the basis of decision-making in AVs, allowing them to perform driving tasks autonomously, efficiently, and more safely than human drivers and offering various economic, social, and environmental benefits. However, algorithmic decision-making in AVs can also introduce new issues that create new safety risks and perpetuate discrimination. We identify bias, ethics, and perverse incentives as key ethical issues in the AV algorithms' decision-making that can create new safety risks and discriminatory outcomes. Technical issues in the AVs' perception, decision-making and control algorithms, limitations of existing AV testing and verification methods, and cybersecurity vulnerabilities can also undermine the performance of the AV system. This article investigates the ethical and technical concerns surrounding algorithmic decision-making in AVs by exploring how driving decisions can perpetuate discrimination and create new safety risks for the public. We discuss steps taken to address these issues, highlight the existing research gaps and the need to mitigate these issues through the design of AV's algorithms and of policies and regulations to fully realise AVs' benefits for smart and sustainable cities.

Meta-learning methods, most notably Model-Agnostic Meta-Learning (Finn et al., 2017) or MAML, have achieved great success in adapting to new tasks quickly, after having been trained on similar tasks. The mechanism behind their success, however, is poorly understood. We begin this work with an experimental analysis of MAML, finding that deep models are crucial for its success, even given sets of simple tasks where a linear model would suffice on any individual task. Furthermore, on image-recognition tasks, we find that the early layers of MAML-trained models learn task-invariant features, while later layers are used for adaptation, providing further evidence that these models require greater capacity than is strictly necessary for their individual tasks. Following our findings, we propose a method which enables better use of model capacity at inference time by separating the adaptation aspect of meta-learning into parameters that are only used for adaptation but are not part of the forward model. We find that our approach enables more effective meta-learning in smaller models, which are suitably sized for the individual tasks. Meta-learning or learning to learn is an appealing notion due to its potential in addressing important challenges when applying machine learning to real-world problems. In particular, learning from prior tasks but being able to to adapt quickly to new tasks improves learning efficiency, model robustness, etc. A promising set of techiques, Model-Agnostic Meta-Learning (Finn et al., 2017) or MAML, and its variants, have received a lot of interest (Nichol et al., 2018; Lee & Choi, 2018; Grant et al., 2018). However, despite several efforts, understanding of how MAML works, either theoretically or in practice, has been lacking (Finn & Levine, 2018; Fallah et al., 2019). For a model that meta-learns, its parameters need to encode not only the common knowledge extracted from the tasks it has seen, which form a task-general inductive bias, but also the capability to adapt to new test tasks (similar to those it has seen) with task-specific knowledge.

This paper studies a rarely explored but critical anomaly detection problem: weakly-supervised anomaly detection with limited labeled anomalies and a large unlabeled data set. This problem is very important because it (i) enables anomaly-informed modeling which helps identify anomalies of interests and address the notorious high false positives in unsupervised anomaly detection, and (ii) eliminates the reliance on large-scale and complete labeled anomaly data in fully-supervised settings. However, the problem is especially challenging since we have only limited labeled data for a single class, and moreover, the seen anomalies often cannot cover all types of anomalies (i.e., unseen anomalies). We address this problem by formulating the problem as a pairwise relation learning task. Particularly, our approach defines a two-stream ordinal regression network to learn the relation of randomly selected instance pairs, i.e., whether the instance pair contains labeled anomalies or just unlabeled data instances. The resulting model leverages both the labeled and unlabeled data to effectively augment the data and learn generalized representations of both normality and abnormality. Extensive empirical results show that our approach (i) significantly outperforms state-of-the-art competing methods in detecting both seen and unseen anomalies and (ii) is substantially more data-efficient. Introduction Anomaly detection aims at identifying exceptional data instances that have a significant deviation from the majority of data instances, which can offer important insights into broad applications, such as identifying fraudulent transactions or insider trading, detecting network intrusions, and early detection of diseases.

We propose a heuristically modified FP-Tree for ontology learning from text. Unlike previous research, for concept extraction, we use a regular expression parser approach widely adopted in compiler construction, i.e., deterministic finite automata (DFA). Thus, the concepts are extracted from unstructured documents. For ontology learning, we use a frequent pattern mining approach and employ a rule mining heuristic function to enhance its quality. This process does not rely on predefined lexico-syntactic patterns, thus, it is applicable for different subjects. We employ the ontology in a question-answering system for students' content-related questions. For validation, we used textbook questions/answers and questions from online course forums. Subject experts rated the quality of the system's answers on a subset of questions and their ratings were used to identify the most appropriate automatic semantic text similarity metric to use as a validation metric for all answers. The Latent Semantic Analysis was identified as the closest to the experts' ratings. We compared the use of our ontology with the use of Text2Onto for the question-answering system and found that with our ontology 80% of the questions were answered, while with Text2Onto only 28.4% were answered, thanks to the finer grained hierarchy our approach is able to produce.