Machine learning plays a role in many deployed decision systems, often in ways that are difficult or impossible to understand by human stakeholders. Explaining, in a human-understandable way, the relationship between the input and output of machine learning models is essential to the development of trustworthy machine-learning-based systems. A burgeoning body of research seeks to define the goals and methods of explainability in machine learning. In this paper, we seek to review and categorize research on counterfactual explanations, a specific class of explanation that provides a link between what could have happened had input to a model been changed in a particular way. Modern approaches to counterfactual explainability in machine learning draw connections to the established legal doctrine in many countries, making them appealing to fielded systems in high-impact areas such as finance and healthcare. Thus, we design a rubric with desirable properties of counterfactual explanation algorithms and comprehensively evaluate all currently-proposed algorithms against that rubric. Our rubric provides easy comparison and comprehension of the advantages and disadvantages of different approaches and serves as an introduction to major research themes in this field. We also identify gaps and discuss promising research directions in the space of counterfactual explainability.

Building models for realistic natural language tasks requires dealing with long texts and accounting for complicated structural dependencies. Neural-symbolic representations have emerged as a way to combine the reasoning capabilities of symbolic methods, with the expressiveness of neural networks. However, most of the existing frameworks for combining neural and symbolic representations have been designed for classic relational learning tasks that work over a universe of symbolic entities and relations. In this paper, we present DRaiL, an open-source declarative framework for specifying deep relational models, designed to support a variety of NLP scenarios. Our framework supports easy integration with expressive language encoders, and provides an interface to study the interactions between representation, inference and learning.

Provenance is a record that describes how entities, activities, and agents have influenced a piece of data. Such provenance information is commonly represented in graphs with relevant labels on both their nodes and edges. With the growing adoption of provenance in a wide range of application domains, increasingly, users are confronted with an abundance of graph data, which may prove challenging to analyse. Graph kernels, on the other hand, have been consistently and successfully used to efficiently classify graphs. In this paper, we introduce a novel graph kernel called \emph{provenance kernel}, which is inspired by and tailored for provenance data. It decomposes a provenance graph into tree-patterns rooted at a given node and considers the labels of edges and nodes up to a certain distance from the root. We employ provenance kernels to classify provenance graphs from three application domains. Our evaluation shows that they perform well in terms of classification accuracy and yield competitive results when compared against standard graph kernel methods and the provenance network analytics method while taking significantly less time.Moreover, we illustrate how the provenance types used in provenance kernels help improve the explainability of predictive models.

He was given extensive access to the secret files of the intelligence agency, although some limits were placed on what he could see and write about, including more recent interceptions of other country's diplomatic messages and some of the technical secrets of code-breaking.

For quite a while, artificial intelligence and machine learning models are leveraged in the healthcare industry to improve patient outcomes.

WASHINGTON (Reuters) - The White House Office of Management and Budget (OMB) said Friday that federal agencies will use artificial intelligence to eliminate outdated, obsolete, and inconsistent requirements across tens of thousands of pages of government regulations. A 2019 pilot project used machine learning algorithms and natural language processing at the Department of Health and Human Services. The test run found hundreds of technical errors and outdated requirements in agency rulebooks, including requests to submit materials by fax. OMB said all federal agencies are being encouraged to update regulations using AI and several agencies have already agreed to do so. Over the last four years, the number of pages in the Code of Federal Regulations has remained at about 185,000.

Researchers from Sandia National Laboratories recently designed machine learning algorithms intended to improve the energy output of nuclear fusion reactors. The research team utilized AI algorithms to simulate the interactions between plasma and materials within the walls of a nuclear fusion reactor. Unlike nuclear fission, which involves splitting atoms apart, the energy created by fusion reactions releases energy through the creation of plasma. Hydrogen atoms are superheated to create a plasma cloud and this cloud releases energy as the particles within it smash into one another and fuse together. This process is chaotic, and if scientists can better control the fusion process, it could lead to substantial increases in the amount of usable energy created by nuclear fusion reactors.

For nearly two years, a small spacecraft called OSIRIS-REx has been orbiting an asteroid more than 100 million miles away, patiently biding its time by studying the rock's surface. Scientists believe that this asteroid, Bennu, is a piece of a much larger one that formed just a few million years after Earth. It's a perfectly preserved cosmic time capsule that could reveal the secrets of the ancient history of our solar system. Tomorrow, OSIRIS-REx will make a daring plunge to Bennu's surface and use a robotic arm to vacuum up some of its space dust, which it'll bring back to Earth. The encounter will last for just a few seconds, but it is a technological feat that has been more than a decade in the making.

"Machine learning [can be used] to identify patterns and spot which vessels stand out as potentially being involved in illegal fishing," Edwards says.

The cyber security threats such as distributed denial-of-service (DDoS) are growing exponentially, disrupting businesses of all sizes globally, leading to outages and loss of user trust, Google has said. The tech giant revealed that its infrastructure absorbed a massive 2.5Tbps DDoS in September 2017, the highest-bandwidth attack reported to date which was the culmination of a six-month campaign that utilised multiple methods of attack. "Despite simultaneously targeting thousands of our IPs, presumably in hopes of slipping past automated defenses, the attack had no impact," Google said in a statement on Friday. The attacker used several networks to spoof 167 Mbps (millions of packets per second) to 180,000 exposed CLDAP, DNS, and SMTP servers, which would then send large responses to Google. "This demonstrates the volumes a well-resourced attacker can achieve: This was four times larger than the record-breaking 623 Gbps attack from the Mirai botnet a year earlier. It remains the highest-bandwidth attack reported to date, leading to reduced confidence in the extrapolation," the company informed.