We introduce a new class of context dependent, incomplete information games to serve as structured prediction models for settings with significant strategic interactions. Our games map the input context to outcomes by first condensing the input into private player types that specify the utilities, weighted interactions, as well as the initial strategies for the players. The game is played over multiple rounds where players respond to weighted aggregates of their neighbors' strategies. The predicted output from the model is a mixed strategy profile (a near-Nash equilibrium) and each observation is thought of as a sample from this strategy profile. We introduce two new aggregator paradigms with provably convergent game dynamics, and characterize the conditions under which our games are identifiable from data. Our games can be parameterized in a transferable manner so that the sets of players can change from one game to another. We demonstrate empirically that our games as models can recover meaningful strategic interactions from real voting data.

In federated learning problems, data is scattered across different servers and exchanging or pooling it is often impractical or prohibited. We develop a Bayesian nonparametric framework for federated learning with neural networks. Each data server is assumed to provide local neural network weights, which are modeled through our framework. We then develop an inference approach that allows us to synthesize a more expressive global network without additional supervision, data pooling and with as few as a single communication round. We then demonstrate the efficacy of our approach on federated learning problems simulated from two popular image classification datasets.

This text discusses several popular explanatory methods that go beyond the error measurements and plots traditionally used to assess machine learning models. Some of the explanatory methods are accepted tools of the trade while others are rigorously derived and backed by long-standing theory. The methods, decision tree surrogate models, individual conditional expectation (ICE) plots, local interpretable model-agnostic explanations (LIME), partial dependence plots, and Shapley explanations, vary in terms of scope, fidelity, and suitable application domain. Along with descriptions of these methods, this text presents real-world usage recommendations supported by a use case and public, in-depth software examples for reproducibility.

On 22 May, the Organization for Economic Co-operation and Development (OECD), an international team working on creating stronger policies in order to improve lives, adopted and approved new Artificial Intelligence (AI) principles. RELATED: WHAT IS EXPLAINABLE ARTIFICIAL INTELLIGENCE AND IS IT NEEDED? OECD principles on AI focus on AI that is original and trustworthy. Respect for human rights and democratic values are also strong focal points of these principles. This is a first of such principles to be agreed upon and put forward by governments.

Artificial intelligence (AI) has been mainly the passion of data science labs and development shops. Lately, however, the implications of its potential impact on business -- in the form of enhanced customer service, expanded intelligent capabilities, and even society at large -- have become clearer. That means the time has come for business leaders to not only understand the implications of AI, but also step up and lead the way. That's because with the great power of AI comes great responsibility. "While AI is quickly becoming a new tool in the CEO tool belt to drive revenues and profitability, it has also become clear that deploying AI requires careful management to prevent unintentional but significant damage, not only to brand reputation but, more important, to workers, individuals, and society as a whole," write Roger Burkhardt, Nicolas Hohn, and Chris Wigley, all with McKinsey.

Was it fair that Harry was hired but not Barry? Was it fair that Pam was fired instead of Sam? How to ensure fairness when an intelligent algorithm takes these decisions instead of a human? How to ensure that the decisions were taken based on merit and not on protected attributes like race or sex? These are the questions that must be answered now that many decisions in real life can be made through machine learning. However research in fairness of algorithms has focused on the counterfactual questions "what if?" or "why?", whereas in real life most subjective questions of consequence are contrastive: "why this but not that?". We introduce concepts and mathematical tools using causal inference to address contrastive fairness in algorithmic decision-making with illustrative thought examples.

When recruiting job candidates, employers rarely observe their underlying skill level directly. Instead, they must administer a series of interviews and/or collate other noisy signals in order to estimate the worker's skill. Traditional economics papers address screening models where employers access worker skill via a single noisy signal. In this paper, we extend this theoretical analysis to a multi-test setting, considering both Bernoulli and Gaussian models. We analyze the optimal employer policy both when the employer sets a fixed number of tests per candidate and when the employer can set a dynamic policy, assigning further tests adaptively based on results from the previous tests. To start, we characterize the optimal policy when employees constitute a single group, demonstrating some interesting trade-offs. Subsequently, we address the multi-group setting, demonstrating that when the noise levels vary across groups, a fundamental impossibility emerges whereby we cannot administer the same number of tests, subject candidates to the same decision rule, and yet realize the same outcomes in both groups.

Machine learning has a great potential to improve data products and business processes. It is used to propose products and news articles that we might be interested in as well as to steer autonomous vehicles and to challenge human experts in non-trivial games. Although machine learning models perform extraordinary well in solving those tasks, we need to be aware of the latent risks that arise through inadvertently encoding bias, responsible for discriminating individuals and strengthening preconceptions, or mistakenly taking random correlation for causation. In her book „Weapons of Math Destruction", Cathy O'Neil even went so far as to say that improvident use of algorithms can perpetuate inequality and threaten democracy. Filter bubbles, racist chat bots, and foolable face detection are prominent examples of malicious outcomes of learning algorithms. With great power comes great responsibility--wise words that every practitioner should keep in mind.

BEIJING - It's the most sensitive day of the year for China's internet, the anniversary of the bloody June 4 crackdown on pro-democracy protests at Tiananmen Square, and with under two weeks to go, China's robot censors are working overtime. Censors at Chinese internet companies say tools to detect and block content related to the 1989 crackdown have reached unprecedented levels of accuracy, aided by machine learning and voice and image recognition. "We sometimes say that the artificial intelligence is a scalpel, and a human is a machete," said one content screening employee at Beijing Bytedance Co., who asked not to be identified because they are not authorized to speak to media. Two employees at the firm said censorship of the Tiananmen crackdown, along with other highly sensitive issues including Taiwan and Tibet, is now largely automated. Posts that allude to dates, images and names associated with the protests are automatically rejected.

In this work we address the problem of argument search. The purpose of argument search is the distillation of pro and contra arguments for requested topics from large text corpora. In previous works, the usual approach is to use a standard search engine to extract text parts which are relevant to the given topic and subsequently use an argument recognition algorithm to select arguments from them. The main challenge in the argument recognition task, which is also known as argument mining, is that often sentences containing arguments are structurally similar to purely informative sentences without any stance about the topic. In fact, they only differ semantically. Most approaches use topic or search term information only for the first search step and therefore assume that arguments can be classified independently of a topic. We argue that topic information is crucial for argument mining, since the topic defines the semantic context of an argument. Precisely, we propose different models for the classification of arguments, which take information about a topic of an argument into account. Moreover, to enrich the context of a topic and to let models understand the context of the potential argument better, we integrate information from different external sources such as Knowledge Graphs or pre-trained NLP models. Our evaluation shows that considering topic information, especially in connection with external information, provides a significant performance boost for the argument mining task.