Papers.

Cloud Computing Providers

Understanding Cloud Computing

Basic Cloud Computing Types

Cloud Computing Evolution

Cloud Computing Basics

Mapping Online Social Media Networks

Detecting and analyzing I/O performance regressions

Deep Speech: Scaling up end-to-end speech recognition

We present a state-of-the-art speech recognition system developed using end-to-end deep learning. Our architecture is significantly simpler than traditional speech systems, which rely on laboriously engineered processing pipelines; these traditional systems also tend to perform poorly when used in noisy environments. In contrast, our system does not need hand-designed components to model background noise, reverberation, or speaker variation, but instead directly learns a function that is robust to such effects. We do not need a phoneme dictionary, nor even the concept of a "phoneme." Key to our approach is a well-optimized RNN training system that uses multiple GPUs, as well as a set of novel data synthesis techniques that allow us to efficiently obtain a large amount of varied data for training. Our system, called Deep Speech, outperforms previously published results on the widely studied Switchboard Hub5'00, achieving 16.0% error on the full test set. Deep Speech also handles challenging noisy environments better than widely used, state-of-the-art commercial speech systems.

3-D Reciprocal Collision Avoidance on Physical Quadrotor Helicopters with On-Board Sensing for Relative Positioning

In this paper, we present an implementation of 3-D reciprocal collision avoidance on real quadrotor helicopters where each quadrotor senses the relative position and velocity of other quadrotors using an on-board camera. We show that using our approach, quadrotors are able to successfully avoid pairwise collisions in GPS and motion-capture denied environments, without communication between the quadrotors, and even when human operators deliberately attempt to induce collisions. To our knowledge, this is the first time that reciprocal collision avoidance has been successfully implemented on real robots where each agent independently observes the others using on-board sensors. We theoretically analyze the response of the collision-avoidance algorithm to the violated assumptions by the use of real robots. We quantitatively analyze our experimental results. A particularly striking observation is that at times the quadrotors exhibit "reciprocal dance" behavior, which is also observed when humans move past each other in constrained environments. This seems to be the result of sensing uncertainty, which causes both robots involved to have a different belief about the relative positions and velocities and, as a result, choose the same side on which to pass.

The Missing Piece in Complex Analytics: Low Latency, Scalable Model Management and Serving with Velox

When is it Better to Compare than to Score?

When eliciting judgements from humans for an unknown quantity, one often has the choice of making direct-scoring (cardinal) or comparative (ordinal) measurements. In this paper we study the relative merits of either choice, providing empirical and theoretical guidelines for the selection of a measurement scheme. We provide empirical evidence based on experiments on Amazon Mechanical Turk that in a variety of tasks, (pairwise-comparative) ordinal measurements have lower per sample noise and are typically faster to elicit than cardinal ones. Ordinal measurements however typically provide less information. We then consider the popular Thurstone and Bradley-Terry-Luce (BTL) models for ordinal measurements and characterize the minimax error rates for estimating the unknown quantity. We compare these minimax error rates to those under cardinal measurement models and quantify for what noise levels ordinal measurements are better. Finally, we revisit the data collected from our experiments and show that fitting these models confirms this prediction: for tasks where the noise in ordinal measurements is sufficiently low, the ordinal approach results in smaller errors in the estimation.

Operational Distributed Regulation for Bitcoin

The fifteen year struggle of decentralizing privacy-enhancing technology

Stochastic Gradient Hamiltonian Monte Carlo

Hamiltonian Monte Carlo (HMC) sampling methods provide a mechanism for defining distant proposals with high acceptance probabilities in a Metropolis-Hastings framework, enabling more efficient exploration of the state space than standard random-walk proposals. The popularity of such methods has grown significantly in recent years. However, a limitation of HMC methods is the required gradient computation for simulation of the Hamiltonian dynamical system--such computation is infeasible in problems involving a large sample size or streaming data. Instead, we must rely on a noisy gradient estimate computed from a subset of the data. In this paper, we explore the properties of such a stochastic gradient HMC approach. Surprisingly, the natural implementation of the stochastic approximation can be arbitrarily bad. To address this problem we introduce a variant that uses second-order Langevin dynamics with a friction term that counteracts the effects of the noisy gradient, maintaining the desired target distribution as the invariant distribution. Results on simulated data validate our theory. We also provide an application of our methods to a classification task using neural networks and to online Bayesian matrix factorization.

GraphX: Unifying Data-Parallel and Graph-Parallel Analytics

From social networks to language modeling, the growing scale and importance of graph data has driven the development of numerous new graph-parallel systems (e.g., Pregel, GraphLab). By restricting the computation that can be expressed and introducing new techniques to partition and distribute the graph, these systems can efficiently execute iterative graph algorithms orders of magnitude faster than more general data-parallel systems. However, the same restrictions that enable the performance gains also make it difficult to express many of the important stages in a typical graph-analytics pipeline: constructing the graph, modifying its structure, or expressing computation that spans multiple graphs. As a consequence, existing graph analytics pipelines compose graph-parallel and data-parallel systems using external storage systems, leading to extensive data movement and complicated programming model. To address these challenges we introduce GraphX, a distributed graph computation framework that unifies graph-parallel and data-parallel computation. GraphX provides a small, core set of graph-parallel operators expressive enough to implement the Pregel and PowerGraph abstractions, yet simple enough to be cast in relational algebra. GraphX uses a collection of query optimization techniques such as automatic join rewrites to efficiently implement these graph-parallel operators. We evaluate GraphX on real-world graphs and workloads and demonstrate that GraphX achieves comparable performance as specialized graph computation systems, while outperforming them in end-to-end graph pipelines. Moreover, GraphX achieves a balance between expressiveness, performance, and ease of use.

Coordination Avoidance in Database Systems (Extended Version)

WOLFE: Strength Reduction and Approximate Programming for Probabilistic Programming

Wish: Amplifying Creative Ability with Expert Crowds

What's In A Name: Precision Medicine and a New Nosology

The Shadow Internet: liberation from Surveillance, Censorship and Servers

The role of venture capital in economic transition in Serbia

The Power of Metaphor in Communicating Risk in Climate Messages

The MemSQL In-Memory Database System.

SystemML's Optimizer: Plan Generation for Large-Scale Machine Learning Programs

Squaring the circle: Managing local healthcare terminologies in the age of standardization

SQL-on-Hadoop: Full Circle Back to Shared-Nothing Database Architectures

SEEDB: Automatically Generating Query Visualizations

Scalable Web Content Understanding Framework

Scala for machine learning : leverage scala and machine learning to construct and study systems that can learn from data

Scala AST Persistence

Robustifying the Sparse Walsh-Hadamard Transform without Increasing the Sample Complexity of O ( K log N )

Reliable, Memory Speed Storage for Cluster Computing Frameworks

ReClaim: a Privacy-Preserving Decentralized Social Network

Recent Advances in Magnetic Tunnel Junction Materials and Stack for Thermally-Assisted Magnetic Random Access Memory

Python

Provable Bounds for Learning Some Deep Representations.

Path Following in Social Web Search

Parallel Double Greedy Submodular Maximization

Many machine learning problems can be reduced to the maximization of sub-modular functions. Although well understood in the serial setting, the parallel maximization of submodular functions remains an open area of research with recent results [1] only addressing monotone functions. The optimal algorithm for maximizing the more general class of non-monotone submodular functions was introduced by Buchbinder et al. [2] and follows a strongly serial double-greedy logic and program analysis. In this work, we propose two methods to parallelize the double-greedy algorithm. The first, coordination-free approach emphasizes speed at the cost of a weaker approximation guarantee. The second, concurrency control approach guarantees a tight 1/2-approximation, at the quantifiable cost of additional coordination and reduced parallelism. As a consequence we explore the tradeoff space between guaranteed performance and objective optimality. We implement and evaluate both algorithms on multi-core hardware and billion edge graphs, demonstrating both the scalability and tradeoffs of each approach. 1

On Communication Cost of Distributed Statistical Estimation and Dimensionality.

More Algorithms for Provable Dictionary Learning.

Making sbt Macro-Aware

Lower Bound for High-Dimensional Statistical Learning Problem via Direct-Sum Theorem.

Incremental latency analysis of heterogeneous cyber-physical systems

REACTION 2014. 3rd International Workshop on Real-time and Distributed Computing in Emerging Applications. Rome, Italy. December 2nd, 2014.

Improving the Performance of Scala Collections with Miniboxing

Hygiene for Scala

The main contribution of this work is a formal model of hygienic name resolution and macro system that is flexible enough to provide missing safety to quasiquotes. This makes it possible to combine the best of two worlds: we get reasonable safety guarantees without sacrificing the notational convenience of quasiquotes.

HTML5 Mobile for Android and iOS

Google Glass Development

Functional Programming in Scala

Enterprise Web Development: Building HTML5 Applications - From Desktop to Mobile

Building enterprise-grade HTML5 web applications that run on desktop and mobile: JavaScript frameworks and tooling, responsive layout, offline storage, WebSockets, testing and deployment, developed through a single sample application.

Encrypted Search & Cluster Formation in Big Data