Yu Liu

With the growing needs of online A/B testing to support the innovation in industry, the opportunity cost of running an experiment becomes non-negligible. Therefore, there is an increasing demand for an efficient continuous monitoring service that allows early stopping when appropriate. Classic statistical methods focus on hypothesis testing and are mostly developed for traditional high-stake problems such as clinical trials, while experiments at online service companies typically have very different features and focuses. Motivated by the real needs, in this paper, we introduce a novel framework that we developed in Amazon to maximize customer experience and control opportunity cost. We formulate the problem as a Bayesian optimal sequential decision making problem that has a unified utility function. We discuss extensively practical design choices and considerations. We further introduce how to solve the optimal decision rule via Reinforcement Learning and scale the solution. We show the effectiveness of this novel approach compared with existing methods via a large-scale meta-analysis on experiments in Amazon.

The forensics discipline of bloodstain pattern analysis plays an important role in crime scene analysis and reconstruction. One reconstruction question is whether the blood has been spattered via gunshot or blunt impact such as beating or stabbing. This paper proposes an automated framework to classify bloodstain spatter patterns generated under controlled conditions into either gunshot or blunt impact classes. Classification is performed using machine learning. The study is performed with 94 blood spatter patterns which are available as public data sets, designs a set of features with possible relevance to classification, and uses the random forests method to rank the most useful features and perform classification. The study shows that classification accuracy decreases with the increasing distance between the target surface collecting the stains and the blood source. Based on the data set used in this study, the model achieves 99% accuracy in classifying spatter patterns at distances of 30 cm, 93% accuracy at distances of 60 cm, and 86% accuracy at distances of 120 cm. Results with 10 additional backspatter patterns also show that the presence of muzzle gases can reduce classification accuracy.