Abhinav Mishra

Zhaohui Wang, Xiao Lin, Abhinav Mishra et al.

Changepoints are abrupt variations in the underlying distribution of data. Detecting changes in a data stream is an important problem with many applications. In this paper, we are interested in changepoint detection algorithms which operate in an online setting in the sense that both its storage requirements and worst-case computational complexity per observation are independent of the number of previous observations. We propose an online changepoint detection algorithm for both univariate and multivariate data which compares favorably with offline changepoint detection algorithms while also operating in a strictly more constrained computational model. In addition, we present a simple online hyperparameter auto tuning technique for these algorithms.

Abhinav Mishra, Ram Sriharsha, Sichen Zhong

Decomposing a complex time series into trend, seasonality, and remainder components is an important primitive that facilitates time series anomaly detection, change point detection, and forecasting. Although numerous batch algorithms are known for time series decomposition, none operate well in an online scalable setting where high throughput and real-time response are paramount. In this paper, we propose OnlineSTL, a novel online algorithm for time series decomposition which is highly scalable and is deployed for real-time metrics monitoring on high-resolution, high-ingest rate data. Experiments on different synthetic and real world time series datasets demonstrate that OnlineSTL achieves orders of magnitude speedups (100x) while maintaining quality of decomposition.

Abhinav Mishra

With increasing importance of e-commerce, many websites have emerged where users can express their opinions about products, such as movies, books, songs, etc. Such interactions can be modeled as bipartite graphs where the weight of the directed edge from a user to a product denotes a rating that the user imparts to the product. These graphs are used for recommendation systems and discovering most reliable (trusted) products. For these applications, it is important to capture the bias of a user when she is rating a product. Users have inherent bias---many users always impart high ratings while many others always rate poorly. It is necessary to know the bias of a reviewer while reading the review of a product. It is equally important to compensate for this bias while assigning a ranking for an object. In this paper, we propose an algorithm to capture the bias of a user and then subdue it to compute the true rating a product deserves. Experiments show the efficiency and effectiveness of our system in capturing the bias of users and then computing the true ratings of a product.