Recently, distant supervision has gained great success on Fine-grained Entity Typing (FET). Despite its efficiency in reducing manual labeling efforts, it also brings the challenge of dealing with false entity type labels, as distant supervision assigns labels in a context agnostic manner. Existing works alleviated this issue with partial-label loss, but usually suffer from confirmation bias, which means the classifier fit a pseudo data distribution given by itself. In this work, we propose to regularize distantly supervised models with Compact Latent Space Clustering (CLSC) to bypass this problem and effectively utilize noisy data yet. Our proposed method first dynamically constructs a similarity graph of different entity mentions; infer the labels of noisy instances via label propagation. Based on the inferred labels, mention embeddings are updated accordingly to encourage entity mentions with close semantics to form a compact cluster in the embedding space,thus leading to better classification performance. Extensive experiments on standard benchmarks show that our CLSC model consistently outperforms state-of-the-art distantly supervised entity typing systems by a significant margin.

Remaining Useful Life (RUL) of an equipment or one of its components is defined as the time left until the equipment or component reaches its end of useful life. Accurate RUL estimation is exceptionally beneficial to Predictive Maintenance, and Prognostics and Health Management (PHM). Data driven approaches which leverage the power of algorithms for RUL estimation using sensor and operational time series data are gaining popularity. Existing algorithms, such as linear regression, Convolutional Neural Network (CNN), Hidden Markov Models (HMMs), and Long Short-Term Memory (LSTM), have their own limitations for the RUL estimation task. In this work, we propose a novel Functional Data Analysis (FDA) method called functional Multilayer Perceptron (functional MLP) for RUL estimation. Functional MLP treats time series data from multiple equipment as a sample of random continuous processes over time. FDA explicitly incorporates both the correlations within the same equipment and the random variations across different equipment's sensor time series into the model. FDA also has the benefit of allowing the relationship between RUL and sensor variables to vary over time. We implement functional MLP on the benchmark NASA C-MAPSS data and evaluate the performance using two popularly-used metrics. Results show the superiority of our algorithm over all the other state-of-the-art methods.

With the wide deployment of machine learning (ML) based systems for a variety of applications including medical, military, automotive, genomic, as well as multimedia and social networking, there is great potential for damage from adversarial learning (AL) attacks. In this paper, we provide a contemporary survey of AL, focused particularly on defenses against attacks on statistical classifiers. After introducing relevant terminology and the goals and range of possible knowledge of both attackers and defenders, we survey recent work on test-time evasion (TTE), data poisoning (DP), and reverse engineering (RE) attacks and particularly defenses against same. In so doing, we distinguish robust classification from anomaly detection (AD), unsupervised from supervised, and statistical hypothesis-based defenses from ones that do not have an explicit null (no attack) hypothesis; we identify the hyperparameters a particular method requires, its computational complexity, as well as the performance measures on which it was evaluated and the obtained quality. We then dig deeper, providing novel insights that challenge conventional AL wisdom and that target unresolved issues, including: 1) robust classification versus AD as a defense strategy; 2) the belief that attack success increases with attack strength, which ignores susceptibility to AD; 3) small perturbations for test-time evasion attacks: a fallacy or a requirement?; 4) validity of the universal assumption that a TTE attacker knows the ground-truth class for the example to be attacked; 5) black, grey, or white box attacks as the standard for defense evaluation; 6) susceptibility of query-based RE to an AD defense. We then present benchmark comparisons of several defenses against TTE, RE, and backdoor DP attacks on images. The paper concludes with a discussion of future work.

Scientist Katherine Bouman has become one of the world's most popular people for helping create the first ever picture of a black hole. The researcher was one of a team made up of a huge number of experts who produced the image, which shows the blazing red and yellow of the event horizon that surrounds the first black hole ever to be seen. And one image in particular of Dr Bouman doing part of that work – using an algorithm she wrote to generate the image that made headlines around the world – has served as a reminder of the vast amount of expertise that has gone into creating such an achievement. We'll tell you what's true. You can form your own view.

Space is doing some strange things to astronauts' bodies, a new study has found. The Nasa twins study is looking at two twins – Scott and Mark Kelly – in an attempt to understand the affects of spaceflight on people's health and physiology. By having one remain on Earth while another lived on the International Space Station for more than a year, researchers are able to compare the two in an attempt to understand what difference to the trip into space made. Now early results show a host of unusual changes happened to Scott when he headed into space. That included the discovery that in some ways, a person's health is not greatly affected at all by a trip into space.

In the new book Fintech, Small Business, & the American Dream, Karen Mills describes how technology is opening up new capital for entrepreneurs. Artificial intelligence, machine learning, and big data will transform financial services and small-business lending long before they impact driverless cars, predicts Harvard Business School Senior Fellow Karen G. Mills. "As we speak, large banks and fintech providers are working on solutions that will dramatically change the options for small businesses." Her new book, Fintech, Small Business, & the American Dream: How Technology is Transforming Lending and Shaping a New Era of Small Business Opportunity, explains why small business is key to the United States economy, and how innovation in financial services will reduce the friction and barriers in small-business lending, helping more of them thrive. In our Q&A, Mills, who served as a member of the Obama Cabinet and headed the US Small Business Administration from 2009 to 2013, outlines the challenges and opportunities facing small businesses, the rise of fintech innovation, and the policy implications for regulation.

Corey Booker and some of his Senate colleagues would like to introduce a new area of government regulation in the tech industry. We need to be keeping a closer eye on the development of Artificial Intelligence, but not because of the coming robot revolution. The problem, you see, is that the computer algorithms are (wait for it)… racist. And that justifies some sort of government oversight of the tech sector beyond what we already have in place today. Congress is starting to show interest in prying open the "black box" of tech companies' artificial intelligence with oversight that parallels how the federal government checks under car hoods and audits banks.

Reprinted with permission from Quanta Magazine's Abstractions blog. In the 1950s, four mathematically minded U.S. Army soldiers used primitive electronic calculators to work out the optimal strategy for playing blackjack. Their results, later published in the Journal of the American Statistical Association, detailed the best decision a player could make for every situation encountered in the game. Yet that strategy--which would evolve into what gamblers call "the book"--did not guarantee a player would win. Blackjack, along with solitaire, checkers, or any number of other games, has a ceiling on the percentage of games in which players can expect to triumph, even if they play the absolute best that the game can be played.

It's well established that algorithms can exhibit bias, however inadvertently, and a trio of US politicians believe they can do something about it. Senators Cory Booker, Yvette Clarke and Ron Wyden have introduced an Algorithmic Accountability Act that would require larger companies to test their algorithms and fix anything "inaccurate, unfair, biased or discriminatory." The move would also ask them to study how their systems protect personal data,and would let the FTC create regulations mandating impact studies for "highly sensitive" automated systems. The bill would only apply to companies that either make more than $50 million per year or have data for at least one million people or devices. Small businesses would theoretically be safe.

It once required an open mind and an active imagination to believe we could launch humans into space. Now, we take human space flight for granted, but we still need that out-of-the-box thinking to push the boundaries of exploration in this solar system and beyond. That's where NASA's Innovative Advanced Concepts (NIAC) program comes in. It's meant to foster ideas that sound borderline science fiction but have the potential to become new technologies. Today, NASA announced 18 innovative concepts that will receive NIAC funding.