Joseph E. Gonzalez

Distributed GraphLab: a framework for machine learning and data mining in the cloud

While high-level data parallel frameworks, like MapReduce, simplify the design and implementation of large-scale data processing systems, they do not naturally or efficiently support many important data mining and machine learning algorithms and can lead to inefficient learning systems. To help fill this critical void, we introduced the GraphLab abstraction which naturally expresses asynchronous, dynamic, graph-parallel computation while ensuring data consistency and achieving a high degree of parallel performance in the shared-memory setting. In this paper, we extend the GraphLab framework to the substantially more challenging distributed setting while preserving strong data consistency guarantees. We develop graph based extensions to pipelined locking and data versioning to reduce network congestion and mitigate the effect of network latency. We also introduce fault tolerance to the GraphLab abstraction using the classic Chandy-Lamport snapshot algorithm and demonstrate how it can be easily implemented by exploiting the GraphLab abstraction itself. Finally, we evaluate our distributed implementation of the GraphLab abstraction on a large Amazon EC2 deployment and show 1-2 orders of magnitude performance gains over Hadoop-based implementations.

GraphX: a resilient distributed graph system on Spark

From social networks to targeted advertising, big graphs capture the structure in data and are central to recent advances in machine learning and data mining. Unfortunately, directly applying existing data-parallel tools to graph computation tasks can be cumbersome and inefficient. The need for intuitive, scalable tools for graph computation has lead to the development of new graph-parallel systems (e.g., Pregel, PowerGraph) which are designed to efficiently execute graph algorithms. Unfortunately, these new graph-parallel systems do not address the challenges of graph construction and transformation which are often just as problematic as the subsequent computation. Furthermore, existing graph-parallel systems provide limited fault-tolerance and support for interactive data mining. We introduce GraphX, which combines the advantages of both data-parallel and graph-parallel systems by efficiently expressing graph computation within the Spark data-parallel framework. We leverage new ideas in distributed graph representation to efficiently distribute graphs as tabular data-structures. Similarly, we leverage advances in data-flow systems to exploit in-memory computation and fault-tolerance. We provide powerful new operations to simplify graph construction and transformation. Using these primitives we implement the PowerGraph and Pregel abstractions in less than 20 lines of code. Finally, by exploiting the Scala foundation of Spark, we enable users to interactively load, transform, and compute on massive graphs.

Scalable inference in latent variable models

Latent variable techniques are pivotal in tasks ranging from predicting user click patterns and targeting ads to organizing the news and managing user generated content. Latent variable techniques like topic modeling, clustering, and subspace estimation provide substantial insight into the latent structure of complex data with little or no external guidance making them ideal for reasoning about large-scale, rapidly evolving datasets. Unfortunately, due to the data dependencies and global state introduced by latent variables and the iterative nature of latent variable inference, latent-variable techniques are often prohibitively expensive to apply to large-scale, streaming datasets. In this paper we present a scalable parallel framework for efficient inference in latent variable models over streaming web-scale data. Our framework addresses three key challenges: 1) synchronizing the global state which includes global latent variables (e.g., cluster centers and dictionaries); 2) efficiently storing and retrieving the large local state which includes the data-points and their corresponding latent variables (e.g., cluster membership); and 3) sequentially incorporating streaming data (e.g., the news). We address these challenges by introducing: 1) a novel delta-based aggregation system with a bandwidth-efficient communication protocol; 2) schedule-aware out-of-core storage; and 3) approximate forward sampling to rapidly incorporate new data. We demonstrate state-of-the-art performance of our framework by easily tackling datasets two orders of magnitude larger than those addressed by the current state-of-the-art. Furthermore, we provide an optimized and easily customizable open-source implementation of the framework1.

Apache Spark: a unified engine for big data processing

This open source computing framework unifies streaming, batch, and interactive big data workloads to unlock new applications.

MLI: An API for Distributed Machine Learning

MLI is an Application Programming Interface designed to address the challenges of building Machine Learning algorithms in a distributed setting based on data-centric computing. Its primary goal is to simplify the development of high-performance, scalable, distributed algorithms. Our initial results show that, relative to existing systems, this interface can be used to build distributed implementations of a wide variety of common Machine Learning algorithms with minimal complexity and highly competitive performance and scalability.

Standards for graph algorithm primitives

It is our view that the state of the art in constructing a large collection of graph algorithms in terms of linear algebraic operations is mature enough to support the emergence of a standard set of primitive building blocks. This paper is a position paper defining the problem and announcing our intention to launch an open effort to define this standard.

Selecting the best VM across multiple public clouds: a data-driven performance modeling approach

Users of cloud services are presented with a bewildering choice of VM types and the choice of VM can have significant implications on performance and cost. In this paper we address the fundamental problem of accurately and economically choosing the best VM for a given workload and user goals. To address the problem of optimal VM selection, we present PARIS, a data-driven system that uses a novel hybrid offline and online data collection and modeling framework to provide accurate performance estimates with minimal data collection. PARIS is able to predict workload performance for different user-specified metrics, and resulting costs for a wide range of VM types and workloads across multiple cloud providers. When compared to sophisticated baselines, including collaborative filtering and a linear interpolation model using measured workload performance on two VM types, PARIS produces significantly better estimates of performance. For instance, it reduces runtime prediction error by a factor of 4 for some workloads on both AWS and Azure. The increased accuracy translates into a 45% reduction in user cost while maintaining performance.

Efficient and accurate clustering for large-scale genetic mapping

High-throughput “next generation” genome sequencing technologies are producing a flood of inexpensive genetic information that is invaluable to genomics research. Sequences of millions of genetic markers are being produced, providing genomics researchers with the opportunity to construct highresolution genetic maps for many complicated genomes. However, the current generation of genetic mapping tools were designed for the small data setting, and are now limited by the prohibitively slow clustering algorithms they employ in the genetic marker-clustering stage. In this work, we present a new approach to genetic mapping based on a fast clustering algorithm that exploits the geometry of the data. Our theoretical and empirical analysis shows that the algorithm can correctly recover linkage groups. Using synthetic and real-world data, including the grand-challenge wheat genome, we demonstrate that our approach can quickly process orders of magnitude more genetic markers than existing tools while retaining - and in some cases even improving - the quality of genetic marker clusters.

From graphs to tables the design of scalable systems for graph analytics

From social networks to language modeling, the growing scale and importance of graph data has driven the development of new graph-parallel systems. In this talk, I will review the graph-parallel abstraction and describe how it can be used to express important machine learning and graph analytics algorithms like PageRank and Latent factor models. I will present how systems like GraphLab and Pregel exploit restrictions in the graph-parallel abstraction along with advances in distributed graph representation to efficiently execute iterative graph algorithms orders of magnitude faster than more general data-parallel systems. Unfortunately, the same restrictions that enable graph-parallel systems to achieve substantial performance gains also limit their ability to express many of the important stages in a typical graph-analytics pipeline. As a consequence, existing approaches to graph-analytics typically compose multiple systems through brittle and costly file interfaces. To fill the need for a holistic approach to graph-analytics we introduce GraphX, which unifies graph-parallel and data-parallel computation under a single API and system. I will show how a simple set of data-parallel operators can be used to express graph-parallel computation and how, by applying a collection of query optimizations derived from our work on graph-parallel systems, we can execute entire graph-analytics pipelines efficiently in a more general data-parallel distributed fault-tolerant system achieving performance comparable to specialized state-of-the-art systems.

SkipNet: Learning Dynamic Routing in Convolutional Networks

While deeper convolutional networks are needed to achieve maximum accuracy in visual perception tasks, for many inputs shallower networks are sufficient. We exploit this observation by learning to skip convolutional layers on a per-input basis. We introduce SkipNet, a modified residual network, that uses a gating network to selectively skip convolutional blocks based on the activations of the previous layer. We formulate the dynamic skipping problem in the context of sequential decision making and propose a hybrid learning algorithm that combines supervised learning and reinforcement learning to address the challenges of non-differentiable skipping decisions. We show SkipNet reduces computation by 30-90% while preserving the accuracy of the original model on four benchmark datasets and outperforms the state-of-the-art dynamic networks and static compression methods. We also qualitatively evaluate the gating policy to reveal a relationship between image scale and saliency and the number of layers skipped.