The Pentagon's top artificial intelligence official warned Tuesday that the department's biggest competitive threat is obsolescence. "The biggest competitive threat is our own obsolescence," said Lt. Gen. Michael Groen, director of the Joint Artificial Intelligence Center. "I could walk out into the parking lot of the Pentagon, turn on my iPhone and join a data-driven, completely integrated environment. I can get whatever services I want. I can review, I can find, I can research. I can do it all at my fingertips. I can't do any of that on a defense network."

We propose a novel interpretable framework for cross-lingual content flagging, which significantly outperforms prior work both in terms of predictive performance and average inference time. The framework is based on a nearest-neighbour architecture and is interpretable by design. Moreover, it can easily adapt to new instances without the need to retrain it from scratch. Unlike prior work, (i) we encode not only the texts, but also the labels in the neighbourhood space (which yields better accuracy), and (ii) we use a bi-encoder instead of a cross-encoder (which saves computation time). Our evaluation results on ten different datasets for abusive language detection in eight languages shows sizable improvements over the state of the art, as well as a speed-up at inference time.

Artificial Intelligence is one of the fastest growing technologies of the 21st century and accompanies us in our daily lives when interacting with technical applications. However, reliance on such technical systems is crucial for their widespread applicability and acceptance. The societal tools to express reliance are usually formalized by lawful regulations, i.e., standards, norms, accreditations, and certificates. Therefore, the T\"UV AUSTRIA Group in cooperation with the Institute for Machine Learning at the Johannes Kepler University Linz, proposes a certification process and an audit catalog for Machine Learning applications. We are convinced that our approach can serve as the foundation for the certification of applications that use Machine Learning and Deep Learning, the techniques that drive the current revolution in Artificial Intelligence. While certain high-risk areas, such as fully autonomous robots in workspaces shared with humans, are still some time away from certification, we aim to cover low-risk applications with our certification procedure. Our holistic approach attempts to analyze Machine Learning applications from multiple perspectives to evaluate and verify the aspects of secure software development, functional requirements, data quality, data protection, and ethics. Inspired by existing work, we introduce four criticality levels to map the criticality of a Machine Learning application regarding the impact of its decisions on people, environment, and organizations. Currently, the audit catalog can be applied to low-risk applications within the scope of supervised learning as commonly encountered in industry. Guided by field experience, scientific developments, and market demands, the audit catalog will be extended and modified accordingly.

The damage and the impact of natural disasters are becoming more destructive with the increase of urbanization. Today's metropolitan cities are not sufficiently prepared for the pre and post-disaster situations. Digital Twin technology can provide a solution. A virtual copy of the physical city could be created by collecting data from sensors of the Internet of Things (IoT) devices and stored on the cloud infrastructure. This virtual copy is kept current and up to date with the continuous flow of the data coming from the sensors. We propose a disaster management system utilizing machine learning called DT-DMS is used to support decision-making mechanisms. This study aims to show how to educate and prepare emergency center staff by simulating potential disaster situations on the virtual copy. The event of a disaster will be simulated allowing emergency center staff to make decisions and depicting the potential outcomes of these decisions. A rescue operation after an earthquake is simulated. Test results are promising and the simulation scope is planned to be extended.

Data transformations (e.g. rotations, reflections, and cropping) play an important role in self-supervised learning. Typically, images are transformed into different views, and neural networks trained on tasks involving these views produce useful feature representations for downstream tasks, including anomaly detection. However, for anomaly detection beyond image data, it is often unclear which transformations to use. Here we present a simple end-to-end procedure for anomaly detection with learnable transformations. The key idea is to embed the transformed data into a semantic space such that the transformed data still resemble their untransformed form, while different transformations are easily distinguishable. Extensive experiments on time series demonstrate that we significantly outperform existing methods on the one-vs.-rest setting but also on the more challenging n-vs.-rest anomaly-detection task. On tabular datasets from the medical and cyber-security domains, our method learns domain-specific transformations and detects anomalies more accurately than previous work.

Attention maps, a popular heatmap-based explanation method for Visual Question Answering (VQA), are supposed to help users understand the model by highlighting portions of the image/question used by the model to infer answers. However, we see that users are often misled by current attention map visualizations that point to relevant regions despite the model producing an incorrect answer. Hence, we propose Error Maps that clarify the error by highlighting image regions where the model is prone to err. Error maps can indicate when a correctly attended region may be processed incorrectly leading to an incorrect answer, and hence, improve users' understanding of those cases. To evaluate our new explanations, we further introduce a metric that simulates users' interpretation of explanations to evaluate their potential helpfulness to understand model correctness. We finally conduct user studies to see that our new explanations help users understand model correctness better than baselines by an expected 30% and that our proxy helpfulness metrics correlate strongly ($\rho$>0.97) with how well users can predict model correctness.

Small-molecule therapeutics treat a wide variety of diseases, but their effectiveness is often diminished because of their pharmacokinetics -- what the body does to a drug. After administration, the body dictates how much of the drug is absorbed, which organs the drug enters, and how quickly the body metabolizes and excretes the drug again. Nanoparticles, usually made out of lipids, polymers, or both, can improve the pharmacokinetics, but they can be complex to produce and often carry very little of the drug. Some combinations of small-molecule cancer drugs and two small-molecule dyes have been shown to self-assemble into nanoparticles with extremely high payloads of drugs, but it is difficult to predict which small-molecule partners will form nanoparticles among the millions of possible pairings. MIT researchers have developed a screening platform that combines machine learning with high-throughput experimentation to identify self-assembling nanoparticles quickly.

Varroa destructor is a deadly stowaway that port authorities are determined to keep away from the bee population in the southeast Australian state of Victoria. Artificially intelligent beehives are being installed at Victorian ports to detect pests as they arrive at ships rapidly. "The Varroa mite is extremely destructive; it kills bees very rapidly," said Mary-Anne Thomas, the Victorian agriculture minister. "I would look forward to a project like the Purple Hive rolling out across the country. Purple Hive was launched on March 29 at the Port of Melbourne -- a solar-powered device that detects Varroa destructor, a mite that feeds on honey bees. Using artificial intelligence and cameras, Purple Hive provides alerts in real-time and has been trialed in New Zealand, where the mite is established. The technology scans each honey bee entering the Purple Hive to determine if Varroa mite is present. The hive is colored purple because it attracts bees. Thomas tweeted a picture of a hive being installed. "At #BegaCheese, we're absolutely buzzing with excitement to announce that B honey's Purple Hive has officially found its first home at the Port of Melbourne, as we join forces with @VicGovAg to help protect honey bee populations from Varroa destructor," read the tweet of Jimmy Coleman, marketing manager of digital and communications, Bega Cheese. "Varroa destructor is the world's most devastating pest of Western honey bees, Apis mellifera Linnaeus," as per the website of the University of Florida. "Accurate estimates of the effect of Varroa on the apiculture industry are hard to find, but it is safe to assume that the mites have killed hundreds of thousands of colonies worldwide, resulting in billions of dollars of economic loss." The adult female mites are reddish-brown to dark brown and oval. Adult males are yellowish with light tan legs and have a spherical body shape. Varroa destructor, the most significant single driver of the global honey bee health decline, was detected on a ship that entered the Port of Melbourne in 2018, but authorities stopped it from becoming an outbreak. "Australia is the only populated country in the world that the Varroa destructor hasn't impacted.

"The beauty of our technology is that it will not only save countless lives but it is incredibly simple," former British Heart Foundation researcher Dr Cheerag Shirodaria, who worked on the project, said.

In practice, real data sets may have missing values or otherwise have only partially observed records that complicate the validity and application validity of standard statistical methodology. Missingness may result from diverse causes, with an underlying mechanism of one of three types: missing completely at random (MCAR), missing at random (MAR), or not missing at random (NMAR) [16]. Under MCAR, the probability that a case (record, sample, observation) is missing feature (variable, attribute, dimension) values does not depend on either the observed or missing feature values. When the probability that a case is missing feature values may depend on the observed feature values, but not the missing feature values, the mechanism is MAR. In the more extreme and challenging case of NMAR, the probability that a case is missing feature values depends on both observed and missing feature values. Notably, if the data are MCAR, they are also MAR; if the data are not MAR, then they are NMAR. Strategies for analysis of data with missing values are often critically dependent on the missingness mechanism, and clustering is no exception. For clustering problems, the most common (and often expedient) treatment of missing values is deletion, on either a case or feature basis, or imputation [17], [18].