We propose a novel algorithm for learning fair representations that can simultaneously mitigate two notions of disparity among different demographic subgroups. Two key components underpinning the design of our algorithm are balanced error rate and conditional alignment of representations. In settings that have historically had discrimination, we are interested in defining fairness with respect to a protected group, the group which has historically been disadvantaged. Among many recent attempts to achieve algorithmic fairness (Dwork et al., 2012; Hardt et al., 2016; Zemel et al., 2013; Zafar et al., 2015), learning fair representations has attracted increasing attention However, it has long been empirically observed (Calders et al., 2009) and recently been proved (Zhao Part of this work was done when Han Zhao was visiting the V ector Institute, Toronto. In this work, we provide an affirmative answer to the above question by proposing an algorithm to align the conditional distributions (on the target variable) of representations across different demographic subgroups.

Antimicrobial resistance (AMR) is the ability of microorganisms like bacteria, viruses, fungi and certain parasites to resist drugs such as antibiotics, antifungals, and antivirals from destroying it. AMR is a worldwide public health threat that is projected to rise. Globally, by 2050, over 10 million deaths per year will be due to antimicrobial resistance according to projections from a report by Wellcome Trust and the UK government. For antibiotic resistance alone, each year over two million people in the U.S. are affected, and 23,000 die, according to figures from the U.S. Centers for Disease Control and Prevention (CDC). Researchers at Washington State University have combined game theory with artificial intelligence (AI) to create a tool that can identify genes that are antibiotic-resistant in bacteria, and published their study in Scientific Reports on October 9, 2019.

Robotic process automation (RPA) uses machine learning (ML) and artificial intelligence (AI) to create a virtual workforce, able to handle repeatable tasks that require a human worker to perform. By using an RPA, companies can perform repetitive tasks faster, longer and with a reduced error rate allowing the workforce to focus on essential duties and responsibilities. In other words, companies have employees working like robots, performing jobs without thinking, why not have robots behaving like people for these tasks. Cybersecurity personnel and cybercriminals are in a constant state of war, automation and specifically RPA can help protect against malicious cyber intruders. Identification and prevention of zero-day attacks (an attack on an exploit the same day of its discovery) and elimination of any system weaknesses is the end goal of internal security teams.

Refining one's hypotheses in the light of data is a commonplace scientific practice, however, this approach introduces selection bias and can lead to specious statistical analysis. One approach of addressing this phenomena is via conditioning on the selection procedure, i.e., how we have used the data to generate our hypotheses, and prevents information to be used again after selection. Many selective inference (a.k.a. post-selection inference) algorithms typically take this approach but will "over-condition" for sake of tractability. While this practice obtains well calibrated $p$-values, it can incur a major loss in power. In our work, we extend two recent proposals for selecting features using the Maximum Mean Discrepancy and Hilbert Schmidt Independence Criterion to condition on the minimal conditioning event. We show how recent advances in multiscale bootstrap makes conditioning on the minimal selection event possible and demonstrate our proposal over a range of synthetic and real world experiments. Our results show that our proposed test is indeed more powerful in most scenarios.

Adversarial training is the standard to train models robust against adversarial examples. However, especially for complex datasets, adversarial training incurs a significant loss in accuracy and is known to generalize poorly to stronger attacks, e.g., larger perturbations or other threat models. In this paper, we introduce confidence-calibrated adversarial training (CCAT) where the key idea is to enforce that the confidence on adversarial examples decays with their distance to the attacked examples. We show that CCAT preserves better the accuracy of normal training while robustness against adversarial examples is achieved via confidence thresholding. Most importantly, in strong contrast to adversarial training, the robustness of CCAT generalizes to larger perturbations and other threat models, not encountered during training. We also discuss our extensive work to design strong adaptive attacks against CCAT and standard adversarial training which is of independent interest. We present experimental results on MNIST, SVHN and Cifar10.

Contextual bandits are online learners that, given an input, select an arm and receive a reward for that arm. They use the reward as a learning signal and aim to maximize the total reward over the inputs. Contextual bandits are commonly used to solve recommendation or ranking problems. This paper considers a learning setting in which multiple parties aim to train a contextual bandit together in a private way: the parties aim to maximize the total reward but do not want to share any of the relevant information they possess with the other parties. Specifically, multiple parties have access to (different) features that may benefit the learner but that cannot be shared with other parties. One of the parties pulls the arm but other parties may not learn which arm was pulled. One party receives the reward but the other parties may not learn the reward value. This paper develops a privacy-preserving contextual bandit algorithm that combines secure multi-party computation with a differential private mechanism based on epsilon-greedy exploration in contextual bandits.

We compare the model-free reinforcement learning with the model-based approaches through the lens of the expressive power of neural networks for policies, $Q$-functions, and dynamics. We show, theoretically and empirically, that even for one-dimensional continuous state space, there are many MDPs whose optimal $Q$-functions and policies are much more complex than the dynamics. We hypothesize many real-world MDPs also have a similar property. For these MDPs, model-based planning is a favorable algorithm, because the resulting policies can approximate the optimal policy significantly better than a neural network parameterization can, and model-free or model-based policy optimization rely on policy parameterization. Motivated by the theory, we apply a simple multi-step model-based bootstrapping planner (BOOTS) to bootstrap a weak $Q$-function into a stronger policy. Empirical results show that applying BOOTS on top of model-based or model-free policy optimization algorithms at the test time improves the performance on MuJoCo benchmark tasks.

How are you using Watson at Nedbank? Traditionally, Nedbank has reduced rates of online fraud by using rule-based decision systems. Every time a fraud was committed, new rules were added. Unfortunately, this created a large catch-all problem where responders would see many false alarms, which would encumber resources and divert attention away from actual fraud. By using machine learning, developed within the Watson Local development environment, we were able to drop the false positive rate from 60–80% to below 35%. Watson helped with easy collaboration for model development and simplified the deployment process by letting us build a production-ready API with ease.

Collecting large-scale data with clean labels for supervised training of neural networks is practically challenging. Although noisy labels are usually cheap to acquire, existing methods suffer severely for training datasets with high noise ratios, making high-cost human labeling a necessity. Here we present a method to train neural networks in a way that is almost invulnerable to severe label noise by utilizing a tiny trusted set. Our method, named IEG, is based on three key insights: (i) Isolation of noisy labels, (ii) Escalation of useful supervision from mislabeled data, and (iii) Guidance from small trusted data. On CIFAR100 with a 40% uniform noise ratio and 10 trusted labeled data per class, our method achieves 80. 2 0.3% classification accuracy, only 1.4% higher error than a neural network trained without label noise. Moreover, increasing the noise ratio to 80%, our method still achieves a high accuracy of 75 .5 Training deep neural networks usually requires large-scale labeled data. However, the process of data labelling by humans is challenging and expensive in practice, especially in domains where expert annotators are needed such as medical imaging. A great number of methods have been proposed to train neural networks from datasets with noisy labels due to cheap acquisition (e.g.

Predicting discomfort glare in open-plan offices is a challenging problem since most of available glare metrics are developed for cellular offices which are typically daylight dominated. The problem with open-plan offices is that they are mainly dependent on electric lighting rather than daylight even when they have a fully glazed facade. In addition, the contrast between bright windows and the buildings interior can be problematic and may cause discomfort glare to the building occupants. These problems can affect occupant productivity and wellbeing. Thus, it is important to develop a predictive model to avoid discomfort glare when designing open plan offices. To the best of our knowledge, we are the first to adopt Machine Learning (ML) models to predict discomfort glare. In order to develop new glare predictive models for these types of offices, Post-Occupancy Evaluation (POE) and High Dynamic Range (HDR) images were collected from 80 occupants (n=80) in four different open-plan offices. Consequently, various multi-region luminance values, luminance and glare indices were calculated and used as input features to train ML models. The accuracy of the ML model was compared to the accuracy of 24 indices which were also evaluated using a Receiver Operating Characteristic (ROC) analysis to identify the best cutoff values (thresholds) for each index for open-plan configurations. Results showed that the ML glare model could predict glare in open-plan offices with an accuracy of 83.8% (0.80 true positive rate and 0.86 true negative rate) which outperformed the accuracy of the previously developed glare metrics.