Consider a nonparametric contextual multi-arm bandit problem where each arm $a \in [K]$ is associated to a nonparametric reward function $f_a: [0,1] \to \mathbb{R}$ mapping from contexts to the expected reward. Suppose that there is a large set of arms, yet there is a simple but unknown structure amongst the arm reward functions, e.g. finite types or smooth with respect to an unknown metric space. We present a novel algorithm which learns data-driven similarities amongst the arms, in order to implement adaptive partitioning of the context-arm space for more efficient learning. We provide regret bounds along with simulations that highlight the algorithm's dependence on the local geometry of the reward functions.

Data science with the kind of power it gives you to analyze each and every bit of data you have at your disposal, to make smart & intelligent business decisions, is becoming a must have tool to understand and implement in your organization, it is very important that your business decisions are not based on intuition rather based on data analysis. "Data which you have in your repository is a gold mine, which needs to be harnessed with an intent to serve the humanity at large, as they are the key source of the same data. Data has a story to tell. Being a data engineer and a business leader it's your primary responsibility to treat them well, process it with appropriate ML model and build a solution which is relevant for both current and future user needs. With this intent, let's begin our journey of understanding supervised ML using Linear Regression model.

We provide path length bounds on gradient descent (GD) and flow (GF) curves for various classes of smooth convex and nonconvex functions. We make six distinct contributions: (a) we prove a meta-theorem that if GD has linear convergence towards an optimal set, then its path length is upper bounded by the distance to the optimal set multiplied by a function of the rate of convergence, (b) under the Polyak-Lojasiewicz (PL) condition (a generalization of strong convexity that allows for certain nonconvex functions), we show that the aforementioned multiplicative factor is at most $\sqrt{\kappa}$, (c) we show an $\widetilde\Omega(\sqrt{d} \wedge \kappa^{1/4})$, times the length of the direct path, lower bound on the worst-case path length for PL functions, (d) for the special case of quadratics, we show that the bound is $\Theta(\min\{\sqrt{d},\sqrt{\log \kappa}\})$ and in some cases can be independent of $\kappa$, (e) under the weaker assumption of just convexity, where there is no natural notion of a condition number, we prove that the path length can be at most $2^{10d^2}$ times the length of the direct path, (f) finally, for separable quasiconvex functions the path length is both upper and lower bounded by ${\Theta}(\sqrt{d})$ times the length of the direct path.

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.

Clustering time series is a delicate task; varying lengths and temporal offsets obscure direct comparisons. A natural strategy is to learn a parametric model foreach time series and to cluster the model parameters rather than the sequences themselves. Linear dynamical systems are a fundamental and powerful parametric model class. However, identifying the parameters of a linear dynamical systems is a venerable task, permitting provably efficient solutions only in special cases. In this work, we show that clustering the parameters of unknown linear dynamical systems is, in fact, easier than identifying them. We analyze a computationally efficient clustering algorithm that enjoys provable convergence guarantees under a natural separation assumption. Although easy to implement, our algorithm is general, handling multi-dimensional data with time offsets and partial sequences. Evaluating our algorithm on both synthetic data and real electrocardiogram (ECG) signals, we see significant improvements in clustering quality over existing baselines.

After admission to emergency department (ED), patients with critical illnesses are transferred to intensive care unit (ICU) due to unexpected clinical deterioration occurrence. Identifying such unplanned ICU transfers is urgently needed for medical physicians to achieve two-fold goals: improving critical care quality and preventing mortality. A priority task is to understand the crucial rationale behind diagnosis results of individual patients during stay in ED, which helps prepare for an early transfer to ICU. Most existing prediction studies were based on univariate analysis or multiple logistic regression to provide one-size-fit-all results. However, patient condition varying from case to case may not be accurately examined by the only judgment. In this study, we present a new decision tool using a mathematical optimization approach aiming to automatically discover rules associating diagnostic features with high-risk outcome (i.e., unplanned transfers) in different deterioration scenarios. We consider four mutually exclusive patient subgroups based on the principal reasons of ED visits: infections, cardiovascular/respiratory diseases, gastrointestinal diseases, and neurological/other diseases at a suburban teaching hospital. The analysis results demonstrate significant rules associated with unplanned transfer outcome for each subgroups and also show comparable prediction accuracy, compared to state-of-the-art machine learning methods while providing easy-to-interpret symptom-outcome information.

In this paper, we present a differential privacy version of convex and nonconvex sparse classification approach. Based on alternating direction method of multiplier (ADMM) algorithm, we transform the solving of sparse problem into the multistep iteration process. Then we add exponential noise to stable steps to achieve privacy protection. By the property of the post-processing holding of differential privacy, the proposed approach satisfies the $\epsilon-$differential privacy even when the original problem is unstable. Furthermore, we present the theoretical privacy bound of the differential privacy classification algorithm. Specifically, the privacy bound of our algorithm is controlled by the algorithm iteration number, the privacy parameter, the parameter of loss function, ADMM pre-selected parameter, and the data size. Finally we apply our framework to logistic regression with $L_1$ regularizer and logistic regression with $L_{1/2}$ regularizer. Numerical studies demonstrate that our method is both effective and efficient which performs well in sensitive data analysis.

In this paper, we consider modelling interaction between a set of variables in the context of time series and high dimension. We suggest two approaches. The first is similar to the neighborhood lasso when the lasso model is replaced by a support vector machine (SVMs). The second is a restricted Bayesian network adapted for time series. We show the efficiency of our approaches by simulations using linear, nonlinear data set and a mixture of both.

Although adaptive gradient methods (AGMs) have fast speed in training deep neural networks, it is known to generalize worse than the stochastic gradient descent (SGD) or SGD with momentum (S-Momentum). Many works have attempted to modify AGMs so to close the gap in generalization performance between AGMs and S-Momentum, but they do not answer why there is such a gap. We identify that the anisotropic scale of the adaptive learning rate (A-LR) used by AGMs contributes to the generalization performance gap, and all existing modified AGMs actually represent efforts in revising the A-LR. Because the A-LR varies significantly across the dimensions of the problem over the optimization epochs (i.e., anisotropic scale), we propose a new AGM by calibrating the A-LR with a {\em softplus} function, resulting in the \textsc{Sadam} and \textsc{SAMSGrad} methods\footnote{Code is available at https://github.com/neilliang90/Sadam.git.}. These methods have better chance to not trap at sharp local minimizers, which helps them resume the dips in the generalization error curve observed with SGD and S-Momentum. We further provide a new way to analyze the convergence of AGMs (e.g., \textsc{Adam}, \textsc{Sadam}, and \textsc{SAMSGrad}) under the nonconvex, non-strongly convex, and Polyak-{\L}ojasiewicz conditions. We prove that the convergence rate of ADAM also depends on its hyper-parameter epsilon, which has been overlooked in prior convergence analysis. Empirical studies support our observation of the anisotropic A-LR and show that the proposed methods outperform existing AGMs and generalize even better than S-Momentum in multiple deep learning tasks.

The full article also features courses that you could attend to learn the topics listed below, as well as numerous comments. We also added a few topics that we think are important and missing in the original article.