Proceedings of the 5th International Workshop on Big Data, Streams and Heterogeneous Source Mining: Algorithms, Systems, Programming Models and Applications at KDD 2016 Held in San Francisco, California, USA on 14 August 2016 Published as Volume 53 by the Proceedings of Machine Learning Research on 06 December 2016. Volume Edited by: Wei Fan Albert Bifet Jesse Read Qiang Yang Philip S. Yu Series Editors: Neil D. Lawrence Mark Reid https://proceedings.mlr.press/v53/ Wed, 08 Feb 2023 10:41:37 +0000 Wed, 08 Feb 2023 10:41:37 +0000 Jekyll v3.9.3 This paper presents a combination of strategies for conversion rate (CVR) prediction de- ployed at the Yahoo! demand side platform (DSP) Brightroll, targeting at modeling extremely high dimensional, sparse data with limited human intervention. We propose a novel probabilistic generative model by tightly integrating components of natural language processing, dynamic transfer learning and scalable prediction, named Dynamic Transfer Learning with Reinforced Word Modeling (a.k.a. Trans-RWM ) to predict user conversion rates. Our model is based on assumptions that: on a higher level, information can be transferable between related campaigns; on a lower level, users who searched similar contents or browsed similar pages would have a higher probability of sharing similar latent purchase interests. Novelties of this framework include (i) A novel natural language modeling specifically tailored for semantic inputs of CVR prediction; (ii) A Bayesian transfer learning model to dynamically transfer the knowledge from source to the future target; (iii) An automatic new updating rule with adaptive regularization using Stochastic Gradient Monte Carlo to support the efficient updating of Trans-RWM in high-dimensional and sparse data. We demonstrate that on Brightroll our framework can effectively discriminate extremely rare events in terms of their conversion propensity. Tue, 06 Dec 2016 00:00:00 +0000 https://proceedings.mlr.press/v53/yang16.html https://proceedings.mlr.press/v53/yang16.html The CDC (Centers for Disease Control and Prevention) currently diagnoses millions of cases of infectious diseases annually, generating population disease distributions that, while accurate, are far too delayed for real-time monitoring. The ability to instantly compile and monitor such distributions is critical in identifying outbreaks and facilitating real-time communication between health authorities and health-care providers. This task, however, is made challenging due to the lack of instantly available public health information, creating a need for the analysis of disease spread on frequently updated social media websites. We introduce a novel pipeline based model to generate a real-time, accurate depiction of infectious disease propagation using Twitter data. Our approach, an amalgam of natural language processing and supervised machine learning, is invariant to mass media hype and significantly reduces the noise introduced by the use of tweets. The correlation coefficient between the Twitter disease distribution obtained via our approach and CDC data from mid-2013 to mid-2014 was 0.983, improving upon the best model published for the 2012-13 flu season. Our model further correlates well with theoretical models of infection spread across airport networks, verifying its robustness and applicability in the public sphere. Tue, 06 Dec 2016 00:00:00 +0000 https://proceedings.mlr.press/v53/shah16.html https://proceedings.mlr.press/v53/shah16.html Support Vector Machines (SVM) are powerful supervised learnings method in machine learning. However, their applicability to large problems, where frequent retraining of the system is required, has been limited due to the time consuming training stage whose computational cost scales quadratically with the number of examples. In this work, a complete FPGA-based system for kernelized SVM training using ensemble learning is presented. The proposed framework builds on the FPGA architecture and utilises a cascaded multiprecision training flow, exploits the heterogeneity within the training problem by tuning the number representation used, and supports ensemble training tuned to each internal memory structure so to address very large datasets. Its performance evaluation shows that the proposed system achieves more than an order of magnitude better results compared to state-of-the-art CPU and GPU-based implementations, providing a stepping stone for researchers and practitioners to tackle large-scale SVM problems that require frequent retraining. Tue, 06 Dec 2016 00:00:00 +0000 https://proceedings.mlr.press/v53/rabieah16.html https://proceedings.mlr.press/v53/rabieah16.html Look-alike audience extension is a practically effective way to customize high-performance audience in on-line advertising. With look-alike audience extension system, any advertiser can easily generate a set of customized audience by just providing a list of existing customers without knowing the detailed targetable attributes in a sophisticated advertising system. In this paper, we present our newly developed graph-based look-alike system in Yahoo! advertising platform which provides look-alike audiences for thousands of campaigns. Extensive experiments have been conducted to compare our look-alike model with three other existing look-alike systems using billions of users and millions of user features. The experiment results show that our developed graph-based method with nearest-neighbor filtering outperforms other methods by more than 50% regarding conversion rate in app-install ad campaigns. Tue, 06 Dec 2016 00:00:00 +0000 https://proceedings.mlr.press/v53/ma16.html https://proceedings.mlr.press/v53/ma16.html Distributed representations of textual elements in low dimensional vector space to capture context has gained great attention recently. Current state-of-the-art word embedding techniques compute distributed representations using co-occurrences of words within a contextual window discounting the flexibility to incorporate other contextual phenomena like temporal, geographical, and topical contexts. In this paper, we present a flexible framework that has the ability to leverage temporal, geographical, and topical information of documents along with the textual content to produce more effective vector representations of entities or words within a document collection. The framework first captures contextual relationships between entities collected from different relevant documents and then leverages these relationships to produce inputs of a graph, or to train a neural network to produce vectors for the entities. Through a set of rigorous experiments we test the performance of our approach and results show that our proposed solution can produce more meaningful vectors than the state-of-the-art methods. Tue, 06 Dec 2016 00:00:00 +0000 https://proceedings.mlr.press/v53/kader16.html https://proceedings.mlr.press/v53/kader16.html Recent years have witnessed a dramatic increase in our ability to collect data from various sensors, devices, in different formats, from independent or connected applications The aim of this workshop is to bring together people from both academia and industry to present their most recent work related to big-data issues, and exchange ideas and thoughts in order to advance this big-data challenge, which has been considered as one of the most exciting opportunities in the past 10 years. Big data flood has outpaced our capability to process, analyze, store and understand these datasets. Consider the Internet data. The web pages indexed by Google were around one million in 1998, but quickly reached 1 billion in 2000 and have already exceeded 1 trillion in 2008. This rapid expansion is accelerated by the dramatic increase in acceptance of social networking applications, such as Facebook, Twitter, Weibo, etc., that allow users to create contents freely and amplify the already huge Web volume. Furthermore, with mobile phones becoming the sensory gateway to get real-time data on people from different aspects, the vast amount of data that mobile carrier can potentially process to improve our daily life has significantly outpaced our past CDR (call data record)-based processing for billing purposes only. It can be foreseen that Internet of things (IoT) applications will raise the scale of data to an unprecedented level. People and devices (from home coffee machines to cars, to buses, railway stations and airports) are all loosely connected. Trillions of such connected components will generate a huge data ocean, and valuable information must be discovered from the data to help improve quality of life and make our world a better place. For example, after we get up every morning, in order to optimize our commute time to work and complete the optimization before we arrive at office, the system needs to process information from traffic, weather, construction, police activities to our calendar schedules, and perform deep optimization under the tight time constraints. In all these applications, we are facing significant challenges in leveraging the vast amount of data, including challenges in (1) system capabilities (2) algorithmic design (3) business models. Tue, 06 Dec 2016 00:00:00 +0000 https://proceedings.mlr.press/v53/fan16.html https://proceedings.mlr.press/v53/fan16.html In this paper, we consider the problem of Bayesian filtering and inference for time series data modeled as noisy, discrete-time observations of a stochastic differential equation (SDE) with undetermined parameters. We develop a Metropolis algorithm to sample from the high-dimensional joint posterior density of all SDE parameters and state time series. Our approach relies on an innovative density tracking by quadrature (DTQ) method to compute the likelihood of the SDE, the part of the posterior that requires the most computational effort to evaluate. As we show, the DTQ method lends itself to a natural implementation using Scala and Apache Spark, an open source framework for scalable data mining. We study the performance and scalability of our algorithm on filtering and inference problems for both regularly and irregularly spaced time series. Tue, 06 Dec 2016 00:00:00 +0000 https://proceedings.mlr.press/v53/bhat16.html https://proceedings.mlr.press/v53/bhat16.html Graphlets represent small induced subgraphs and are becoming increasingly important for a variety of applications. Despite the importance of the local graphlet problem, existing work focuses mainly on counting graphlets globally over the entire graph. These global counts have been used for tasks such as graph classification as well as for understanding and summarizing the fundamental structural patterns in graphs. In contrast, this work proposes a flexible, efficient, and scalable parallel framework for the more challenging problem of counting graphlets locally for a given edge or set of edges.The local graphlet counts provide a topologically rigorous characterization of the local structure surrounding an edge. The aim of this work is to obtain the count of every graphlet of size k ∈{3, 4 } for each edge. The framework gives rise to efficient, parallel, and accurate unbiased estimation methods as well as exact graphlet algorithms for counting graphlets locally. Experiments demonstrate the effectiveness of the proposed exact and estimation methods. Tue, 06 Dec 2016 00:00:00 +0000 https://proceedings.mlr.press/v53/ahmed16.html https://proceedings.mlr.press/v53/ahmed16.html