Explaining the output of a complex machine learning (ML) model often requires approximation using a simpler model. To construct interpretable explanations that are also consistent with the original ML model, counterfactual examples --- showing how the model's output changes with small perturbations to the input --- have been proposed. This paper extends the work in counterfactual explanations by addressing the challenge of feasibility of such examples. For explanations of ML models in critical domains such as healthcare, finance, etc, counterfactual examples are useful for an end-user only to the extent that perturbation of feature inputs is feasible in the real world. We formulate the problem of feasibility as preserving causal relationships among input features and present a method that uses (partial) structural causal models to generate actionable counterfactuals. When feasibility constraints may not be easily expressed, we propose an alternative method that optimizes for feasibility as people interact with its output and provide oracle-like feedback. Our experiments on a Bayesian network and the widely used "Adult" dataset show that our proposed methods can generate counterfactual explanations that satisfy feasibility constraints.

With advances in Artificial Intelligence in Education (AIEd) and the ever-growing scale of Interactive Educational Systems (IESs), data-driven approach has become a common recipe for various tasks such as knowledge tracing and learning path recommendation. Unfortunately, collecting real students' interaction data is often challenging, which results in the lack of public large-scale benchmark dataset reflecting a wide variety of student behaviors in modern IESs. Although several datasets, such as ASSISTments, Junyi Academy, Synthetic and STATICS, are publicly available and widely used, they are not large enough to leverage the full potential of state-of-the-art data-driven models and limits the recorded behaviors to question-solving activities. To this end, we introduce EdNet, a large-scale hierarchical dataset of diverse student activities collected by Santa, a multi-platform self-study solution equipped with artificial intelligence tutoring system. EdNet contains 131,441,538 interactions from 784,309 students collected over more than 2 years, which is the largest among the ITS datasets released to the public so far. Unlike existing datasets, EdNet provides a wide variety of student actions ranging from question-solving to lecture consumption and item purchasing. Also, EdNet has a hierarchical structure where the student actions are divided into 4 different levels of abstractions. The features of EdNet are domain-agnostic, allowing EdNet to be extended to different domains easily. The dataset is publicly released under Creative Commons Attribution-NonCommercial 4.0 International license for research purposes. We plan to host challenges in multiple AIEd tasks with EdNet to provide a common ground for the fair comparison between different state of the art models and encourage the development of practical and effective methods.

One of the key approaches to save samples when learning a policy for a reinforcement learning problem is to use knowledge from an approximate model such as its simulator. However, does knowledge transfer from approximate models always help to learn a better policy? Despite numerous empirical studies of transfer reinforcement learning, an answer to this question is still elusive. In this paper, we provide a strong negative result, showing that even the full knowledge of an approximate model may not help reduce the number of samples for learning an accurate policy of the true model. We construct an example of reinforcement learning models and show that the complexity with or without knowledge transfer has the same order. On the bright side, effective knowledge transferring is still possible under additional assumptions. In particular, we demonstrate that knowing the (linear) bases of the true model significantly reduces the number of samples for learning an accurate policy.

Differential privacy provides a rigorous framework to quantify data privacy, and has received considerable interest recently. A randomized mechanism satisfying $(\epsilon, \delta)$-differential privacy (DP) roughly means that, except with a small probability $\delta$, altering a record in a dataset cannot change the probability that an output is seen by more than a multiplicative factor $e^{\epsilon} $. A well-known solution to $(\epsilon, \delta)$-DP is the Gaussian mechanism initiated by Dwork et al. [1] in 2006 with an improvement by Dwork and Roth [2] in 2014, where a Gaussian noise amount $\sqrt{2\ln \frac{2}{\delta}} \times \frac{\Delta}{\epsilon}$ of [1] or $\sqrt{2\ln \frac{1.25}{\delta}} \times \frac{\Delta}{\epsilon}$ of [2] is added independently to each dimension of the query result, for a query with $\ell_2$-sensitivity $\Delta$. Although both classical Gaussian mechanisms [1,2] assume $0 < \epsilon \leq 1$, our review finds that many studies in the literature have used the classical Gaussian mechanisms under values of $\epsilon$ and $\delta$ where the added noise amounts of [1,2] do not achieve $(\epsilon,\delta)$-DP. We obtain such result by analyzing the optimal noise amount $\sigma_{DP-OPT}$ for $(\epsilon,\delta)$-DP and identifying $\epsilon$ and $\delta$ where the noise amounts of classical mechanisms are even less than $\sigma_{DP-OPT}$. Since $\sigma_{DP-OPT}$ has no closed-form expression and needs to be approximated in an iterative manner, we propose Gaussian mechanisms by deriving closed-form upper bounds for $\sigma_{DP-OPT}$. Our mechanisms achieve $(\epsilon,\delta)$-DP for any $\epsilon$, while the classical mechanisms [1,2] do not achieve $(\epsilon,\delta)$-DP for large $\epsilon$ given $\delta$. Moreover, the utilities of our mechanisms improve those of [1,2] and are close to that of the optimal yet more computationally expensive Gaussian mechanism.

Today at AWS re:Invent in Las Vegas, NFL commissioner Roger Goodell joined AWS CEO Andy Jassy on stage to announce a new partnership to use machine learning to help reduce head injuries in professional football. "We're excited to announce a new strategic partnership together, which is going to combine cloud computing, machine learning and data science to work on transforming player health and safety," Jassy said today. NFL football is a fast and violent sport involving large men. Injuries are a part of the game, but the NFL is hoping to reduce head injuries in particular, a huge problem for the sport. A 2017 study found that 110 out of 111 deceased NFL players had chronic traumatic encephalopathy (CTE).

On Sunday, New South Wales began rolling out a system of cameras designed to detect drivers using their phones illegally. The goal: make the Australian state's roads safer. "Some people have not got the message about using their phones legally and safely," New South Wales Minister for Roads Andrew Constance said in a news release. "If they think they can continue to put the safety of themselves, their passengers, and the community at risk without consequence, they are in for a rude shock." The cameras snap photos of drivers and then use artificial intelligence to determine whether the driver was using a mobile phone illegally.

At the Current Trends in Biotherapeutics Workshop, Alexandre Le Bouthillier, PhD, Co-Founder, Imagia, presented as part of Session 3: Future Trends in Translational Medicine. His talk was called "Personalized Healthcare with Artificial Intelligence." The Clinical Translation Education Group (CTEG) hosted its 2019 workshop on current trends and innovations in cell and gene therapy with an emphasis on disruptive technologies. The session introduced new tools and strategies that are shaping where the biotherapeutic field is headed. The workshop took place in Toronto on September 29, 2019.

Global payments unicorn Adyen is looking into how artificial intelligence can boost its payments offering, highlighting how the tech can boost the industry as a whole. The Dutch FinTech company went public last year after having raised a massive $250m Series B round in 2014, the biggest deal raised by a FinTech company in the Netherlands between 2014 and the third quarter of 2019. But more is to come and it seems as if AI will play a big role in Adyen's future. "The benefits of AI are real," Pieter van der Does, CEP of Adyen, told VentureBeat at the Slush technology conference in Helsinki. However, initially the tech leader was cautious about using AI.

Lord Drayson seems a man on the move. As an amateur racing driver, it is perhaps an innate charateristic, and in the current debate around health data his foot is very much on the gas. I meet the former Labour government science and defence minister shortly after he lays out a bold new vision for the NHS at FutureScot's recent Digital Health & Care conference in Glasgow. Urbane and well-connected, Drayson is also a keen student of policy and how the arguments around big tech and health data are shaping up. For background, there is an intensifying argument that the NHS needs to make much more use of a still largely untapped goldmine of data, which could herald a new dawn in the way we diagnose, treat and manage disease – not to mention save billions of pounds annually.

Artificial intelligence and AI are transformative advancements that level many playing fields, such a large number of in reality that a small country can militarily contend with extraordinary military power, similar to the US. The Chinese have an open, exceptionally profound, amazingly well-subsidized pledge to AI. Aviation based armed forces General VeraLinn Jamieson says it evidently: "We gauge the complete spending on artificial intelligence frameworks in China in 2017 was $12 billion. We likewise gauge that it will develop to at any rate $70 billion by 2020." Andrew Yang, during a Democratic Candidates banter, expressed that the US is losing the AI weapons contest to China. Barely a year back, I contended something very similar.