The original problem of supervised classification considers the task of automatically assigning objects to their respective classes on the basis of numerical measurements derived from these objects. Classifiers are the tools that implement the actual functional mapping from these measurements---also called features or inputs---to the so-called class label---or output. The fields of pattern recognition and machine learning study ways of constructing such classifiers. The main idea behind supervised methods is that of learning from examples: given a number of example input-output relations, to what extent can the general mapping be learned that takes any new and unseen feature vector to its correct class? This chapter provides a basic introduction to the underlying ideas of how to come to a supervised classification problem. In addition, it provides an overview of some specific classification techniques, delves into the issues of object representation and classifier evaluation, and (very) briefly covers some variations on the basic supervised classification task that may also be of interest to the practitioner.

Building classification models is an intrinsically practical exercise that requires many design decisions prior to deployment. We aim to provide some guidance in this decision making process. Specifically, given a classification problem with real valued attributes, we consider which classifier or family of classifiers should one use. Strong contenders are tree based homogeneous ensembles, support vector machines or deep neural networks. All three families of model could claim to be state-of-the-art, and yet it is not clear when one is preferable to the others. Our extensive experiments with over 200 data sets from two distinct archives demonstrate that, rather than choose a single family and expend computing resources on optimising that model, it is significantly better to build simpler versions of classifiers from each family and ensemble. We show that the Heterogeneous Ensembles of Standard Classification Algorithms (HESCA), which ensembles based on error estimates formed on the train data, is significantly better (in terms of error, balanced error, negative log likelihood and area under the ROC curve) than its individual components, picking the component that is best on train data, and a support vector machine tuned over 1089 different parameter configurations. We demonstrate HESCA+, which contains a deep neural network, a support vector machine and two decision tree forests, is significantly better than its components, picking the best component, and HESCA. We analyse the results further and find that HESCA and HESCA+ are of particular value when the train set size is relatively small and the problem has multiple classes. HESCA is a fast approach that is, on average, as good as state-of-the-art classifiers, whereas HESCA+ is significantly better than average and represents a strong benchmark for future research.

In this paper we present a technique for using the bootstrap to estimate the operating characteristics and their variability for certain types of ensemble methods. Bootstrapping a model can require a huge amount of work if the training data set is large. Fortunately in many cases the technique lets us determine the effect of infinite resampling without actually refitting a single model. We apply the technique to the study of meta-parameter selection for random forests. We demonstrate that alternatives to bootstrap aggregation and to considering \sqrt{d} features to split each node, where d is the number of features, can produce improvements in predictive accuracy.

Binary classification is highly used in credit scoring in the estimation of probability of default. The validation of such predictive models is based both on rank ability, and also on calibration (i.e. how accurately the probabilities output by the model map to the observed probabilities). In this study we cover the current best practices regarding calibration for binary classification, and explore how different approaches yield different results on real world credit scoring data. The limitations of evaluating credit scoring models using only rank ability metrics are explored. A benchmark is run on 18 real world datasets, and results compared. The calibration techniques used are Platt Scaling and Isotonic Regression. Also, different machine learning models are used: Logistic Regression, Random Forest Classifiers, and Gradient Boosting Classifiers. Results show that when the dataset is treated as a time series, the use of re-calibration with Isotonic Regression is able to improve the long term calibration better than the alternative methods. Using re-calibration, the non-parametric models are able to outperform the Logistic Regression on Brier Score Loss.

We propose a novel adaptive test of goodness-of-fit, with computational cost linear in the number of samples. We learn the test features that best indicate the differences between observed samples and a reference model, by minimizing the false negative rate. These features are constructed via Stein's method, meaning that it is not necessary to compute the normalising constant of the model. We analyse the asymptotic Bahadur efficiency of the new test, and prove that under a mean-shift alternative, our test always has greater relative efficiency than a previous linear-time kernel test, regardless of the choice of parameters for that test. In experiments, the performance of our method exceeds that of the earlier linear-time test, and matches or exceeds the power of a quadratic-time kernel test. In high dimensions and where model structure may be exploited, our goodness of fit test performs far better than a quadratic-time two-sample test based on the Maximum Mean Discrepancy, with samples drawn from the model.

In this work, we addressed the issue of applying a stochastic classifier and a local, fuzzy confusion matrix under the framework of multi-label classification. We proposed a novel solution to the problem of correcting label pairwise ensembles. The main step of the correction procedure is to compute classifier- specific competence and cross-competence measures, which estimates error pattern of the underlying classifier. We considered two improvements of the method of obtaining confusion matrices. The first one is aimed to deal with imbalanced labels. The other utilizes double labelled instances which are usually removed during the pairwise transformation. The proposed methods were evaluated using 29 benchmark datasets. In order to assess the efficiency of the introduced models, they were compared against 1 state-of-the-art approach and the correction scheme based on the original method of confusion matrix estimation. The comparison was performed using four different multi-label evaluation measures: macro and micro-averaged F1 loss, zero-one loss and Hamming loss. Additionally, we investigated relations between classification quality, which is expressed in terms of different quality criteria, and characteristics of multi-label datasets such as average imbalance ratio or label density. The experimental study reveals that the correction approaches significantly outperforms the reference method only in terms of zero-one loss.

These question can make you think THRICE! Machine learning and data science are being looked as the drivers of the next industrial revolution happening in the world today. This also means that there are numerous exciting startups looking for data scientists. What could be a better start for your aspiring career! However, still, getting into these roles is not easy. You obviously need to get excited about the idea, team and the vision of the company. You might also find some real difficult techincal questions on your way. The set of questions asked depend on what does the startup do. Do they build ML products? You should always find this out prior to beginning your interview preparation. To help you prepare for your next interview, I've prepared a list of 40 plausible & tricky questions which are likely to come across your way in interviews. If you can answer and understand these question, rest assured, you will give a tough fight in your job interview. Note: A key to answer these questions is to have concrete practical understanding on ML and related statistical concepts.

The growing trend of using wearable devices for context-aware computing and pervasive sensing systems has raised its potentials for quick and reliable authentication techniques. Since personal writing habitats differ from each other, it is possible to realize user authentication through writing. This is of great significance as sensible information is easily collected by these devices. This paper presents a novel user authentication system through wrist-worn devices by analyzing the interaction behavior with users, which is both accurate and efficient for future usage. The key feature of our approach lies in using much more effective Savitzky-Golay filter and Dynamic Time Wrapping method to obtain fine-grained writing metrics for user authentication. These new metrics are relatively unique from person to person and independent of the computing platform. Analyses are conducted on the wristband-interaction data collected from 50 users with diversity in gender, age, and height. Extensive experimental results show that the proposed approach can identify users in a timely and accurate manner, with a false-negative rate of 1.78\%, false-positive rate of 6.7\%, and Area Under ROC Curve of 0.983 . Additional examination on robustness to various mimic attacks, tolerance to training data, and comparisons to further analyze the applicability.

In supervised learning, given a sample of n pairs of inputs and outputs, the goal is to estimate a function to be used to predict future outputs based on observing only the corresponding inputs. The quality of an estimate is often measured in terms of the mean-squared prediction error, in which case the regression function is optimal. Since the properties of the function to be estimated are not known a priori, nonparametric techniques, that can adapt their complexity to the problem at hand, are often key to good results. Kernel methods [15, 36] are probably the most common nonparametric approaches to learning. They are based on choosing a reproducing kernel Hilbert space (RKHS) as the hypothesis space in the design of learning algorithms. A classical learning algorithm using kernel methods to perform learning tasks is kernel ridge regression (KRR), which is based on minimizing the sum of a data-fitting term and an explicit penalty term. The penalty term is used for regularization, and controls the complexity of the solution, preventing overfitting. The statistical properties of KRR have been studied extensively, see e.g.

It's no secret that one of Google's strengths in recent years has been voice recognition. In my own experience, my Google Home picks up what I am trying to say almost every time, even in a low voice. Obviously the success rate varies by language and accent, but it is still pretty darn impressive. At Google I/O 2017, Google revealed that the error rate for the company's speech recognition technology has fallen to just under 5% for US English. This is a 3.6% drop from this time last year.