With the industry trend of shifting from a traditional hierarchical approach to flatter management structure, crowdsourced performance assessment gained mainstream popularity. One fundamental challenge of crowdsourced performance assessment is the risks that personal interest can introduce distortions of facts, especially when the system is used to determine merit pay or promotion. In this paper, we developed a method to identify bias and strategic behavior in crowdsourced performance assessment, using a rich dataset collected from a professional service firm in China. We find a pattern of "discriminatory generosity" on the part of peer evaluation, where raters downgrade their peer coworkers who have passed objective promotion requirements while overrating their peer coworkers who have not yet passed. This introduces two types of biases: the first aimed against more competent competitors, and the other favoring less eligible peers which can serve as a mask of the first bias. This paper also aims to bring angles of fairness-aware data mining to talent and management computing. Historical decision records, such as performance ratings, often contain subjective judgment which is prone to bias and strategic behavior. For practitioners of predictive talent analytics, it is important to investigate potential bias and strategic behavior underlying historical decision records.

They are ubiquitous in both scientific and business documents. Data visualizations are designed to be effective at conveying trends and comparisons in a glance, while also preserving salient details. Using computer vision to parse these visualizations can enable extraction of information that cannot be gleaned by solely studying a document's text. Despite the high potential payoff and tremendous practical value, this problem has received little attention until recently. In 2018, two datasets for answering questions about data visualizations were introduced along with new algorithms [15, 18]; however, there is considerable room for improvement. Here, we propose a novel algorithm that exceeds the state-of-the-art on both of these datasets by a large margin. Visual question answering (VQA) requires a system to answer questions about images [6, 27, 17].

We compare several approaches to learn an Optimal Map, represented as a neural network, between probability distributions. The approaches fall into two categories: ``Heuristics'' and approaches with a more sound mathematical justification, motivated by the dual of the Kantorovitch problem. Among the algorithms we consider a novel approach involving dynamic flows and reductions of Optimal Transport to supervised learning.

China is betting big on the potential of artificial intelligence to revolutionize education. A newly published MIT Technology Review story details how the nation is embracing AI as both a replacement and a supplement to human teachers -- and the outcome of the country's AI experiment could affect the future of education on a global scale. From algorithms that curate tutoring lessons to surveillance systems that monitor classroom progress, tens of millions of Chinese students currently rely on some sort of AI to help them learn, MIT Tech reports, with three elements factoring into AI-powered education's ability to thrive in China. For one, the nation has made it a point to incentivize such efforts through tax breaks. Then there's the fact that education is already something of a competitive sport in China, with students -- and their parents -- willing to try anything that might increase their test scores even slightly. Finally, the people developing these AIs have a wealth of data available for training purposes as China places less of an emphasis on individual data privacy than many other developed countries.

We all have to read this - just to see how many different ways there are to solve the same problem. It's the world's biggest experiment on #AIineducation, and no one can predict the outcome. Silicon Valley is also keenly interested. In a report in March, the Chan-Zuckerberg Initiative and the Bill and Melinda Gates Foundation identified AI as an educational tool worthy of investment. China is undergoing the largest-scale experiment on artificial intelligence in education. Here's what's happening and how it could shape the rest of the world.

Ada Lovelace was an English mathematician who lived in the first half of the 19th century. In 1842, Lovelace was tasked with translating an article from French into English for Charles Babbage, the "Grandfather of the Computer." Babbage's piece was about his Analytical Engine, a revolutionary new automatic calculating machine. Although originally retained solely to translate the article, Lovelace also scribbled extensive ideas about the machine into the margins, adding her unique insight, seeing that the Analytical Engine could be used to decode symbols and to make music, art, and graphics. Her notes, which included a method for calculating the Bernoulli numbers sequence and for what would become known as the "Lovelace objection," were the first computer programs on record, even though the machine could not actually be built at the time.1 Though never formally trained as a mathematician, Lovelace was able to see beyond the limitations of Babbage's invention and imagine the power and potential of programmable computers; also, she was a woman, and women in the first half of the 19th century were typically not seen as suited for this type of career. Lovelace had to sign her work with just her initials because women weren't thought of as proper authors at the time.2 Still, she persevered,3 and her work, which would eventually be considered the world's first computer algorithm, later earned her the title of the first computer programmer.

In the last couple of years, Uber has quietly become one of the most active contributors to open source machine learning technologies. From training frameworks like Horovod, statistical languages like Pyro or conversational stacks like the Plato Research Dialogue System, Uber has been pushing boundaries of innovation in the machine learning space with practical technologies rather than exoteric research. One Uber's most popular contributions to the machine learning ecosystem has been Ludwig, a framework for training and testing machine learning models without the need to write code. Recently, Uber released a second version of Ludwig that includes major enhancements in order to enable mainstream no-code experiences for machine learning developers. The goal of Ludwig is to simplify the processes of training and testing machine learning models using a declarative, no-code experience.

Due to the high demand in computation and memory, deep learning solutions are mostly restricted to high-performance computing units, e.g., those present in servers, Cloud, and computing centers. In pervasive systems, e.g., those involving Internet-of-Things (IoT) technological solutions, this would require the transmission of acquired data from IoT sensors to the computing platform and wait for its output. This solution might become infeasible when remote connectivity is either unavailable or limited in bandwidth. Moreover, it introduces uncertainty in the "data production to decision making"-latency, which, in turn, might impair control loop stability if the response should be used to drive IoT actuators. In order to support a real-time recall phase directly at the IoT level, deep learning solutions must be completely rethought having in mind the constraints on memory and computation characterizing IoT units. In this paper we focus on Convolutional Neural Networks (CNNs), a specific deep learning solution for image and video classification, and introduce a methodology aiming at distributing their computation onto the units of the IoT system. We formalize such a methodology as an optimization problem where the latency between the data-gathering phase and the subsequent decision-making one is minimized. The methodology supports multiple IoT sources of data as well as multiple CNNs in execution on the same IoT system, making it a general-purpose distributed computing platform for CNN-based applications demanding autonomy, low decision-latency, and high Quality-of-Service.

We study the relationship between catastrophic forgetting and properties of task sequences. In particular, given a sequence of tasks, we would like to understand which properties of this sequence influence the error rates of continual learning algorithms trained on the sequence. To this end, we propose a new procedure that makes use of recent developments in task space modeling as well as correlation analysis to specify and analyze the properties we are interested in. As an application, we apply our procedure to study two properties of a task sequence: (1) total complexity and (2) sequential heterogeneity. We show that error rates are strongly and positively correlated to a task sequence's total complexity for some state-of-the-art algorithms. We also show that, surprisingly, the error rates have no or even negative correlations in some cases to sequential heterogeneity. Our findings suggest directions for improving continual learning benchmarks and methods.