Much of biomedical and healthcare data is encoded in discrete, symbolic form such as text and medical codes. There is a wealth of expert-curated biomedical domain knowledge stored in knowledge bases and ontologies, but the lack of reliable methods for learning knowledge representation has limited their usefulness in machine learning applications. While text-based representation learning has significantly improved in recent years through advances in natural language processing, attempts to learn biomedical concept embeddings so far have been lacking. A recent family of models called knowledge graph embeddings have shown promising results on general domain knowledge graphs, and we explore their capabilities in the biomedical domain. We train several state-of-the-art knowledge graph embedding models on the SNOMED-CT knowledge graph, provide a benchmark with comparison to existing methods and in-depth discussion on best practices, and make a case for the importance of leveraging the multi-relational nature of knowledge graphs for learning biomedical knowledge representation. The embeddings, code, and materials will be made available to the communitY.

For many tasks, the reward function is too complex to be specified procedurally, and must instead be learned from user data. Prior work has evaluated learned reward functions by examining rollouts from a policy optimized for the learned reward. However, this method cannot distinguish between the learned reward function failing to reflect user preferences, and the reinforcement learning algorithm failing to optimize the learned reward. Moreover, the rollout method is highly sensitive to details of the environment the learned reward is evaluated in, which often differ in the deployment environment. To address these problems, we introduce the Equivalent-Policy Invariant Comparison (EPIC) distance to quantify the difference between two reward functions directly, without training a policy. We prove EPIC is invariant on an equivalence class of reward functions that always induce the same optimal policy. Furthermore, we find EPIC can be precisely approximated and is more robust than baselines to the choice of visitation distribution. Finally, we find that the EPIC distance of learned reward functions to the ground-truth reward is predictive of the success of training a policy, even in different transition dynamics.

In data sequences measured over space or time, an important problem is accurate detection of abrupt changes. In partially labeled data, it is important to correctly predict presence/absence of changes in positive/negative labeled regions, in both the train and test sets. One existing dynamic programming algorithm is designed for prediction in unlabeled test regions (and ignores the labels in the train set); another is for accurate fitting of train labels (but does not predict changepoints in unlabeled test regions). We resolve these issues by proposing a new optimal changepoint detection model that is guaranteed to fit the labels in the train data, and can also provide predictions of unlabeled changepoints in test data. We propose a new dynamic programming algorithm, Labeled Optimal Partitioning (LOPART), and we provide a formal proof that it solves the resulting non-convex optimization problem. We provide theoretical and empirical analysis of the time complexity of our algorithm, in terms of the number of labels and the size of the data sequence to segment. Finally, we provide empirical evidence that our algorithm is more accurate than the existing baselines, in terms of train and test label error.

We derive a family of loss functions to train models in the presence of sampling bias. Examples are when the prevalence of a pathology differs from its sampling rate in the training dataset, or when a machine learning practioner rebalances their training dataset. Sampling bias causes large discrepancies between model performance in the lab and in more realistic settings. It is omnipresent in medical imaging applications, yet is often overlooked at training time or addressed on an ad-hoc basis. Our approach is based on Bayesian risk minimization. For arbitrary likelihood models we derive the associated bias corrected loss for training, exhibiting a direct connection to information gain. The approach integrates seamlessly in the current paradigm of (deep) learning using stochastic backpropagation and naturally with Bayesian models. We illustrate the methodology on case studies of lung nodule malignancy grading.

Web-scale applications can ship code on a daily to weekly cadence. These applications rely on online metrics to monitor the health of new releases. Regressions in metric values need to be detected and diagnosed as early as possible to reduce the disruption to users and product owners. Regressions in metrics can surface due to a variety of reasons: genuine product regressions, changes in user population, and bias due to telemetry loss (or processing) are among the common causes. Diagnosing the cause of these metric regressions is costly for engineering teams as they need to invest time in finding the root cause of the issue as soon as possible. We present Lumos, a Python library built using the principles of AB testing to systematically diagnose metric regressions to automate such analysis. Lumos has been deployed across the component teams in Microsoft's Real-Time Communication applications Skype and Microsoft Teams. It has enabled engineering teams to detect 100s of real changes in metrics and reject 1000s of false alarms detected by anomaly detectors. The application of Lumos has resulted in freeing up as much as 95% of the time allocated to metric-based investigations. In this work, we open source Lumos and present our results from applying it to two different components within the RTC group over millions of sessions. This general library can be coupled with any production system to manage the volume of alerting efficiently.

We observe that a crucial inventory management problem ('phantom inventory'), that by some measures costs retailers approximately 4% in annual sales can be viewed as a problem of identifying anomalies in a (low-rank) Poisson matrix. State of the art approaches to anomaly detection in low-rank matrices apparently fall short. Specifically, from a theoretical perspective, recovery guarantees for these approaches require that non-anomalous entries be observed with vanishingly small noise (which is not the case in our problem, and indeed in many applications). So motivated, we propose a conceptually simple entry-wise approach to anomaly detection in low-rank Poisson matrices. Our approach accommodates a general class of probabilistic anomaly models. We extend recent work on entry-wise error guarantees for matrix completion, establishing such guarantees for sub-exponential matrices, where in addition to missing entries, a fraction of entries are corrupted by (an also unknown) anomaly model. We show that for any given budget on the false positive rate (FPR), our approach achieves a true positive rate (TPR) that approaches the TPR of an (unachievable) optimal algorithm at a min-max optimal rate. Using data from a massive consumer goods retailer, we show that our approach provides significant improvements over incumbent approaches to anomaly detection.

In the classification of a class imbalance dataset, the performance measure used for the model selection and comparison to competing methods is a major issue. In order to overcome this problem several performance measures are defined and analyzed in several perspectives regarding in particular the imbalance ratio. There is still no clear indication which metric is universal and can be used for any skewed data problem. In this paper we introduced a new performance measure based on the harmonic mean of Recall and Selectivity normalized in class labels. This paper shows that the proposed performance measure has the right properties for the imbalanced dataset. In particular, in the space defined by the majority class examples and imbalance ratio it is less sensitive to changes in the majority class and more sensitive to changes in the minority class compared with other existing single-value performance measures. Additionally, the identity of the other performance measures has been proven analytically.

Staartjes have contributed equally to this series, and share first authorship. Abstract As analytical machine learning tools become readily available for clinicians to use, the understanding of key concepts and the awareness of analytical pitfalls are increasingly required for clinicians, investigators, reviewers and editors, who even as experts in their clinical field, sometimes find themselves insufficiently equipped to evaluate machine learning methodologies. In this section, we provide explanations on the general principles of machine learning, as well as analytical steps required for successful machine learning-based predictive modelling, which is the focus of this series. In particular, we define the terms machine learning, artificial intelligence, as well as supervised and unsupervised learning, continuing by introducing optimization, thus, the minimization of an objective error function as the central dogma of machine learning. In addition, we discuss why it is important to separate predictive and explanatory modelling, and most importantly state that a prediction model should not be used to make inferences. Lastly, we broadly describe a classical workflow for training a machine learning model, starting with data pre-processing and feature engineering and selection, continuing on with a training structure consisting of a resampling method, hyperparameter tuning, and model selection, and ending with evaluation of model discrimination and calibration as well as robust internal or external validation of the fully developed model. Methodological rigor and clarity as well as understanding of the underlying reasoning of the internal workings of a machine learning approach are required, otherwise predictive applications despite being strong analytical tools are not well accepted into the clinical routine.

Counterfactual explanations are usually obtained by identifying the smallest change made to an input to change a prediction made by a fixed model (hereafter called sparse methods). Recent work, however, has revitalized an old insight: there often does not exist one superior solution to a prediction problem with respect to commonly used measures of interest (e.g. error rate). In fact, often multiple different classifiers give almost equal solutions. This phenomenon is known as predictive multiplicity (Breiman, 2001; Marx et al., 2019). In this work, we derive a general upper bound for the costs of counterfactual explanations under predictive multiplicity. Most notably, it depends on a discrepancy notion between two classifiers, which describes how differently they treat negatively predicted individuals. We then compare sparse and data support approaches empirically on real-world data. The results show that data support methods are more robust to multiplicity of different models. At the same time, we show that those methods have provably higher cost of generating counterfactual explanations under one fixed model. In summary, our theoretical and empiricaln results challenge the commonly held view that counterfactual recommendations should be sparse in general.

With the abundance of machine learning methods available and the temptation of using them all in an ensemble method, having a model-agnostic method of feature selection is incredibly alluring. Principal component analysis was developed in 1901 and has been a strong contender in this role since, but in the end is an unsupervised method. It offers no guarantee that the features that are selected have good predictive power because it does not know what is being predicted. To this end, Peng et al. developed the minimum redundancy-maximum relevance (mRMR) method in 2005. It uses the mutual information not only between predictors but also includes the mutual information with the response in its calculation. Estimating mutual information and entropy tend to be expensive and problematic endeavors, which leads to excessive processing times even for dataset that is approximately 750 by 750 in a Leave-One-Subject-Out jackknife situation. To remedy this, we use a method from 2012 called Distance Correlation Sure Independence Screening (DC-SIS) which uses the distance correlation measure of Sz\'ekely et al. to select features that have the greatest dependence with the response. We show that this method produces statistically indistinguishable results to the mRMR selection method on Parkinson's Disease vocal diagnosis data 90 times faster.