This paper revisits the offline change-point detection problem from a statistical learning perspective. Instead of assuming that the underlying pre-and post-change distributions are known, it is assumed that we have partial knowledge of these distributions based on empirically observed statistics in the form of training sequences. Our problem formulation finds a variety of real-life applications from detecting when climate change occurred to detecting when a virus mutated. Using the training sequences as well as the test sequence consisting of a single-change and allowing for the erasure or rejection option, we derive the optimal resolution between the estimated and true change-points under two different asymptotic regimes on the undetected error probability--namely, the large and moderate deviations regimes. In both regimes, strong converses are also proved. In the moderate deviations case, the optimal resolution is a simple function of a symmetrized version of the chi-square distance. I. INTRODUCTION AND MOTIVATION The change-point detection (CPD) problem consists in finding changes in the underlying statistical model of data sequences that are modelled as time series. This problem has a plethora of applications in industrial systems [1], medical diagnoses [2], environmental monitoring [3], speech processing [4], finance, economics, and so on [5]. The CPD problems can be divided into two main types: offline CPD and online CPD [6]; the latter is also known as sequential CPD. This depends on whether the data sequence is fixed or obtained in a real-time setting. Offline CPD is a problem that is studied in, for example, anomaly detection problems such as detecting climate change based on existing and known statistics.

We introduce AutoGluon-Tabular, an open-source AutoML framework that requires only a single line of Python to train highly accurate machine learning models on an unprocessed tabular dataset such as a CSV file. Unlike existing AutoML frameworks that primarily focus on model/hyperparameter selection, AutoGluon-Tabular succeeds by ensembling multiple models and stacking them in multiple layers. Experiments reveal that our multi-layer combination of many models offers better use of allocated training time than seeking out the best. A second contribution is an extensive evaluation of public and commercial AutoML platforms including TPOT, H2O, AutoWEKA, auto-sklearn, AutoGluon, and Google AutoML Tables. Tests on a suite of 50 classification and regression tasks from Kaggle and the OpenML AutoML Benchmark reveal that AutoGluon is faster, more robust, and much more accurate. We find that AutoGluon often even outperforms the best-in-hindsight combination of all of its competitors. In two popular Kaggle competitions, AutoGluon beat 99% of the participating data scientists after merely 4h of training on the raw data.

The damage caused by crypto-ransomware, due to encryption, is difficult to revert and cause data losses. In this paper, a machine learning (ML) classifier was built to early detect ransomware (called crypto-ransomware) that uses cryptography by program behavior. If a signature-based detection was missed, a behavior-based detector can be the last line of defense to detect and contain the damages. We find that input/output activities of ransomware and the file-content entropy are unique traits to detect crypto-ransomware. A deep-learning (DL) classifier can detect ransomware with a high accuracy and a low false positive rate. We conduct an adversarial research against the models generated. We use simulated ransomware programs to launch a gray-box analysis to probe the weakness of ML classifiers and to improve model robustness. In addition to accuracy and resiliency, trustworthiness is the other key criteria for a quality detector. Making sure that the correct information was used for inference is important for a security application. The Integrated Gradient method was used to explain the deep learning model and also to reveal why false negatives evade the detection. The approaches to build and to evaluate a real-world detector were demonstrated and discussed.

The increasing scale of distributed learning problems necessitates the development of compression techniques for reducing the information exchange between compute nodes. The level of accuracy in existing compression techniques is typically chosen before training, meaning that they are unlikely to adapt well to the problems that they are solving without extensive hyper-parameter tuning. In this paper, we propose dynamic tuning rules that adapt to the communicated gradients at each iteration. In particular, our rules optimize the communication efficiency at each iteration by maximizing the improvement in the objective function that is achieved per communicated bit. Our theoretical results and experiments indicate that the automatic tuning strategies significantly increase communication efficiency on several state-of-the-art compression schemes.

We study the invariance properties of alignment in linear neural networks under gradient descent. Alignment of weight matrices is a form of implicit regularization, and previous works have studied this phenomenon in fully connected networks with 1-dimensional outputs. In such networks, we prove that there exists an initialization such that adjacent layers remain aligned throughout training under any real-valued loss function. We then define alignment for fully connected networks with multidimensional outputs and prove that it generally cannot be an invariant for such networks under the squared loss. Moreover, we characterize the datasets under which alignment is possible. We then analyze networks with layer constraints such as convolutional networks. In particular, we prove that gradient descent is equivalent to projected gradient descent, and show that alignment is impossible given sufficiently large datasets. Importantly, since our definition of alignment is a relaxation of balancedness, our negative results extend to this property.

The notion of implicit bias, or implicit regularization, has been suggested as a means to explain the surprising generalization ability of modern-days overparameterized learning algorithms. This notion refers to the tendency of the optimization algorithm towards a certain structured solution that often generalizes well. Recently, several papers have studied implicit regularization and were able to identify this phenomenon in various scenarios. We revisit this paradigm in arguably the simplest non-trivial setup, and study the implicit bias of Stochastic Gradient Descent (SGD) in the context of Stochastic Convex Optimization. As a first step, we provide a simple construction that rules out the existence of a \emph{distribution-independent} implicit regularizer that governs the generalization ability of SGD. We then demonstrate a learning problem that rules out a very general class of \emph{distribution-dependent} implicit regularizers from explaining generalization, which includes strongly convex regularizers as well as non-degenerate norm-based regularizations. Certain aspects of our constructions point out to significant difficulties in providing a comprehensive explanation of an algorithm's generalization performance by solely arguing about its implicit regularization properties.

Artificial intelligence (AI) systems, powered by massive data and sophisticated algorithms -- including but not limited to -- deep neural networks and statistical machine learning (ML)(support vector machines, clustering, random forest, etc.), are having profound and transformative impact on our daily lives as they make their way into everything from finance to healthcare, from retail to transportation. Netflix movie recommender, Amazon's product prediction, Facebook's uncanny ability to show what you may like, Google's assistant, DeepMind's AlphaGo, Stanford's AI beating human doctors. Machine learning is eating software. However, one of the common features of these powerful algorithms is that they utilize sophisticated mathematics to do their job -- to classify and segment an image, to arrive at the key decisions, to make a product recommendation, to model a complex phenomenon, or to extract and visualize a hidden pattern from a deluge of data. All of these mathematical processes are, quite simply, beyond the scope of a single human (or a team) to perform manually (even on a computer) or inside their head.

Data can tell us stories. That's what I've been told anyway. As a Data Scientist working for Fortune 300 clients, I deal with tons of data daily, I can tell you that data can tell us stories. You can apply a regression, classification or a clustering algorithm on the data, but feature selection and engineering can be a daunting task. A lot of times, I have seen data scientists take an automated approach to feature selection such as Recursive Feature Elimination (RFE) or leverage Feature Importance algorithms using Random Forest or XGBoost.

In recent years, post-hoc local instance-level and global dataset-level explainability of black-box models has received a lot of attention. Much less attention has been given to obtaining insights at intermediate or group levels, which is a need outlined in recent works that study the challenges in realizing the guidelines in the General Data Protection Regulation (GDPR). In this paper, we propose a meta-method that, given a typical local explainability method, can build a multilevel explanation tree. The leaves of this tree correspond to the local explanations, the root corresponds to the global explanation, and intermediate levels correspond to explanations for groups of data points that it automatically clusters. The method can also leverage side information, where users can specify points for which they may want the explanations to be similar. We argue that such a multilevel structure can also be an effective form of communication, where one could obtain few explanations that characterize the entire dataset by considering an appropriate level in our explanation tree. Explanations for novel test points can be cost-efficiently obtained by associating them with the closest training points. When the local explainability technique is generalized additive (viz. LIME, GAMs), we develop a fast approximate algorithm for building the multilevel tree and study its convergence behavior. We validate the effectiveness of the proposed technique based on two human studies -- one with experts and the other with non-expert users -- on real world datasets, and show that we produce high fidelity sparse explanations on several other public datasets.

One of the core challenges in the field of control theory and reinforcement learning is adaptive control. It is the problem of controlling dynamical systems when the dynamics of the systems are unknown to the decision-making agents. In adaptive control, agents interact with given systems in order to explore and control them while the long-term objective is to minimize the overall average associated costs. The agent has to balance between exploration and exploitation, learn the dynamics, strategize for further exploration, and exploit the estimation to minimize the overall costs. The sequential nature of agent-system interaction results in challenges in the system identifying, estimation, and control under uncertainty, and these challenges are magnified when the systems are partially observable, i.e. contain hidden underlying dynamics. In the linear systems, when the underlying dynamics are fully observable, the asymptotic optimality of estimation methods has been the topic of study in the last decades [Lai et al., 1982, Lai and Wei, 1987]. Recently, novel techniques and learning algorithms have been developed to study the finite-time behavior of adaptive control algorithms and shed light on the design of optimal methods [Peña et al., 2009, Fiechter, 1997, Abbasi-Yadkori and Szepesvári, 2011]. In particular, Abbasi-Yadkori and Szepesvári [2011] proposes to use the principle of optimism in the face of uncertainty (OFU) to balance exploration and exploitation in LQR, where the state of the system is observable.