Publications

Shuffling, Fast and Slow: Scalable Analytics on Serverless Infrastructure

Serverless computing is poised to fulfill the long-held promise of transparent elasticity and millisecond-level pricing. To achieve this goal, service providers impose a fine-grained computational model where every

USENIX Symposium on Networked Systems Design and Implementation (NSDI'19),
Author(s): Qifan Pu, Shivaram Venkataraman, Ion Stoica

ALICE: Autonomous Link-based Cell Scheduling for TSCH

Although low-power lossy network (LLN), at its early stage, commonly used asynchronous link layer protocols for simple operation on resource-constrained nodes, development of embedded hardware and time synchronizat

IPSN 2019: The 18th ACM/IEEE Conference on Information Processing in Sensor Networks,
Author(s): Seohyang Kim, Hyung-Sin Kim, Chongkwon Kim

Context: The Missing Piece in the Machine Learning Lifecycle

Machine learning models have become ubiquitous in modern applications. The ML Lifecycle describes a three-phase process used by data scientists and data engineers to develop, train, and serve models.

Workshop on Common Model Infrastructure, KDD 2018,
Author(s): Rolando Garcia, Vikram Sreekanti, Neeraja Yadwadkar, Dan Crankshaw, Joseph Gonzalez, Joe Hellerstein

Confluo: Distributed Monitoring and Diagnosis Stack for High-speed Networks

Confluo is an end-host stack that can be integrated with existing network management tools to enable monitoring and diagnosis of network-wide events using telemetry data distributed across end-hosts, even for

USENIX Symposium on Networked Systems Design and Implementation (NSDI'19), https://people.eecs.berkeley.edu/~anuragk/papers/confluo.pdf
Author(s): Anurag Khandelwal, Rachit Agarwal, Ion Stoica

FairFuzz: A Targeted Mutation Strategy for Increasing Greybox Fuzz Testing Coverage

In recent years, fuzz testing has proven itself to be one of the most effective techniques for finding correctness bugs and security vulnerabilities in practice. One particular fuzz testing tool, American Fuzzy Lop

33rd ACM/IEEE International Conference on Automated Software Engineering, https://dl.acm.org/citation.cfm?id=3213874
Author(s): Caroline Lemieux, Koushik Sen

PerfFuzz: Automatically Generating Pathological Inputs

Performance problems in software can arise unexpectedly when programs are provided with inputs that exhibit worst-case behavior. A large body of work has focused on diagnosing such problems via statistical profilin

27th International Symposium on Software Testing and Analysis, https://dl.acm.org/citation.cfm?id=3213874
Author(s): Caroline Lemieux, Rohan Padhye, Koushik Sen, Dawn song

Benchmarks for reinforcement learning in mixed-autonomy traffic

We release new benchmarks in the use of deep reinforcement learning (RL) to create controllers for mixed-autonomy traffic, where connected and autonomous vehicles (CAVs) interact with human drivers and infrastruct

2nd Conference on Robot Learning, http://proceedings.mlr.press/v87/vinitsky18a/vinitsky18a.pdf
Author(s): Eugene Vinitsky, Aboudy Kriedieh, Luc Le Flem, Nishant Kheterpal, Kathy Jang, Cathy Wu, Richard Liaw, Eric Liang, Alexandre Bayen

Mitigating the Latency-Accuracy Trade-off in Mobile Data Analytics Systems

An increasing amount of mobile analytics is performed on data that is procured in a real-time fashion to make real-time decisions. Such tasks include simple reporting on streams to sophisticated model building. How

24th Annual International Conference on Mobile Computing and Networking (MobiCom),
Author(s): Anand Padmanabha Iyer, Li Erran Li, Mosharaf Chowdhury, Ion Stoica

e-mission: An Open-Source, Smartphone Platform for Collecting Human Travel Data

GPS-equipped smartphones provide new methods to collect data about travel behavior, including travel survey apps that incorporate automated location sensing. Previous approaches to this have involved proprietary or

Transportation Research Record: Journal of the Transportation Research Board, https://doi.org/10.1177/0361198118770167
Author(s): K. Shankari, Mohamed Amine Bouzaghrane, Samuel M. Maurer, Paul Waddell, David Culler, Randy Katz

ASAP: Fast, Approximate Graph Pattern Mining at Scale

While there has been a tremendous interest in processing data that has an underlying graph structure, existing distributed graph processing systems take several minutes or even hours to mine simple patterns on grap

USENIX Symposium on Operating Systems Design and Implementation (OSDI), https://www.usenix.org/conference/osdi18/presentation/iyer
Author(s): Anand Padmanabha Iyer, Zaoxing Liu, Xin Jin, Shivaram Venkataraman, Vladimir Braverman, Ion Stoica

MARVEL: Enabling Mobile Augmented Reality with Low Energy and Low Latency

This paper presents MARVEL, a mobile augmented reality (MAR) system which provides a notation display service with imperceptible latency (<100 ms) and low energy consumption on regular mobile devices. In contrast t

The 16th ACM Conferenceon Embedded Networked Sensor Systems (SenSys ’18),
Author(s): Kaifei Chen, Tong Li, Hyung-Sin Kim, David Culler, Randy Katz

System Architecture Directions for Post-SoC/32-bit Networked Sensors

The emergence of low-power 32-bit Systems-on-Chip (SoCs), which integrate a 32-bit MCU, radio, and flash, presents an opportunity to re-examine design points and trade-offs at all levels of the system architecture

The 16th ACM Conference on Embedded Networked Sensor Systems (SenSys ’18),
Author(s): Hyung-Sin Kim, Michael Andersen, Kaifei Chen, Sam Kumar, William J. Zhao, Kevin Ma, David Culler

Tributary: spot-dancing for elastic services with latency SLOs

The Tributary elastic control system embraces the uncertain nature of transient cloud resources, such as AWS spot instances, to manage elastic services with latency SLOs more robustly and more cost-effectively. Suc

2018 USENIX Annual Technical Conference, https://www.usenix.org/conference/atc18/presentation/harlap
Author(s): Aaron Harlap, Andrew Chung, Alexey Tumanov, Gregory R. Ganger, Phillip B. Gibbons

3Sigma: distribution-based cluster scheduling for runtime uncertainty

The 3Sigma cluster scheduling system uses job runtime histories in a new way. Knowing how long each job will execute enables a scheduler to more effectively pack jobs with diverse time concerns (e.g., deadline vs.

ACM European Conference on Computer Systems (EuroSys'2018), https://dl.acm.org/citation.cfm?id=3190515
Author(s): Jun Woo Park, Alexey Tumanov, Angela Jiang, Michael A. Kozuch, Gregory R. Ganger

Parametrized Hierarchical Procedures for Neural Programming

Parametrized Hierarchical Procedures (PHP) represent a program as a hierarchy of procedures that call each other, each implemented by a neural network. We develop an algorithm for training PHPs from a set of

International Conferences on Learning Representations (ICLR) 2018, https://openreview.net/pdf?id=rJl63fZRb
Author(s): Roy Fox, Eui Chul (Richard) Shin, Sanjay Krishnan, Kenneth Goldberg, Dawn song, Ion Stoica

FireSim: FPGA-Accelerated Cycle-Exact Scale-Out System Simulation in the Public Cloud

We present FireSim, an open-source simulation platform that enables cycle-exact microarchitectural simulation of large scale-out clusters by combining FPGA-accelerated simulation of silicon-proven RTL designs

2018 ACM/IEEE 45th Annual International Symposium on Computer Architecture (ISCA 2018), https://sagark.org/assets/pubs/firesim-isca2018.pdf
Author(s): Sagar Karandikar, Howard Mao, Donggyu Kim, David Biancolin, Alon Amid, Dayeol Lee, Nathan Pemberton, Emmanuel Amaro, Colin Schmidt, Aditya Chopra, Qijing Huang, Kyle Kovacs, Borivoje Nikolic, Randy Katz, Jonathan Bachrach, Krste Asanovic

Towards Specification-Directed Program Repair

Several recent papers have developed neural network program synthesizers by using supervised learning over large sets of randomly generated programs and specifications. In this paper, we investigate the

International Conferences on Learning Representations (ICLR) 2018, workshop track, https://openreview.net/pdf?id=B1iZRFkwz
Author(s): Eui Chul (Richard) Shin, Illia Polosukhin, Dawn song

Tree-to-tree Neural Networks for Program Translation

Program translation is an important tool to migrate legacy code in one language into an ecosystem built in a different language. In this work, we are the first to consider employing deep neural networks toward tack

International Conferences on Learning Representations (ICLR) 2018, workshop track, https://openreview.net/pdf?id=rkxY-sl0W
Author(s): Xinyun Chen, Chang Liu, Dawn song

Towards Synthesizing Complex Programs From Input-Output Examples

In recent years, deep learning techniques have been developed to improve the performance of program synthesis from input-output examples. Albeit its significant progress, the programs that can be synthesized by sta

International Conferences on Learning Representations (ICLR) 2018, https://openreview.net/pdf?id=Skp1ESxRZ
Author(s): Xinyun Chen, Chang Liu, Dawn song

SkipNet: Learning Dynamic Routing in Convolutional Networks

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.

European Conference on Computer Vision (ECCV) 2018, https://arxiv.org/abs/1711.09485
Author(s): Xin Wang, Fisher Yu, Zi-Yi Dou, Trevor Darrell, Joseph Gonzalez

IDK Cascades: Fast Deep Learning by Learning not to Overthink

We introduce the "I Don't Know"(IDK) prediction cascades framework, a general framework to systematically compose a set of pre-trained models to accelerate inference without a loss in prediction accuracy.

Conference on Uncertainty in Artificial Intelligence (UAI) 2018, http://auai.org/uai2018/proceedings/papers/212.pdf
Author(s): Xin Wang, Yujia Luo, Dan Crankshaw, Alexey Tumanov, Fisher Yu, Joseph Gonzalez

Do Not Lose Bandwidth: Adaptive Transmission Power and Multihop Topology Control

We show that a multihop wireless network can achieve better bandwidth and routing stability when transmission power and routing topology are jointly and adaptively controlled. Our experiments show that the

13th International Conference on Distributed Computing in Sensor Systems (DCOSS 2017),
Author(s): Hyung-Sin Kim, Jeongyeup Paek, David Culler, Saewoong Bahk

Challenging the IPv6 Routing Protocol for Low-Power and Lossy Networks (RPL): A Survey

RPL is the IPv6 routing protocol for low-power and lossy networks, standardized by IETF in 2012 as RFC6550. Specifically, RPL is designed to be a simple and inter-operable networking protocol for resource-constrain

IEEE Communications Surveys & Tutorials ( Volume: 19, Issue: 4, Fourthquarter 2017 ),
Author(s): Hyung-Sin Kim, Jeonggil Ko, David Culler, Jeongyeup Paek

SnapLink: Fast and Accurate Vision-Based Appliance Control in Large Commercial Buildings.

As the number and heterogeneity of appliances in smart buildings increases, identifying and controlling them becomes challenging. Existing methods face various challenges when deployed in large commercial buildings

Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, Vol. 1, No. 4, https://people.eecs.berkeley.edu/~kaifei/data/ubicomp_18.pdf
Author(s): Kaifei Chen, Jonathan Fürst, Jack Kolb, Hyung-Sin Kim, Xin Jin, David Culler, Randy Katz

Towards Fast and Scalable Graph Pattern Mining

While there has been a tremendous interest in processing graph-structured data, existing distributed graph processing systems take several minutes or even hours to mine simple patterns on graphs. In this paper, we

10th USENIX Workshop on Hot Topics in Cloud Computing (HotCloud '18), https://www.usenix.org/conference/hotcloud18/presentation/iyer-pattern
Author(s): Anand Padmanabha Iyer, Zaoxing Liu, Xin Jin, Shivaram Venkataraman, Vladimir Braverman, Ion Stoica

Monarch: Gaining Command on Geo-Distributed Graph Analytics

A number of existing and emerging application scenarios generate graph-structured data in a geo-distributed fashion. Although there is a lot of interest in distributed graph processing systems, none of them support

10th USENIX Workshop on Hot Topics in Cloud Computing (HotCloud '18), https://www.usenix.org/conference/hotcloud18/presentation/iyer-monarch
Author(s): Anand Padmanabha Iyer, Aurojit Panda, Mosharaf Chowdhury, Aditya Akella, Scott Shenker, Ion Stoica

Bridging the GAP: Towards Approximate Graph Analytics

SIGMOD Graph Data-management Experiences & Systems (GRADES), https://dl.acm.org/citation.cfm?id=3210269
Author(s): Anand Padmanabha Iyer, Aurojit Panda, Shivaram Venkataraman, Mosharaf Chowdhury, Aditya Akella, Scott Shenker, Ion Stoica

DIZK: A Distributed Zero Knowledge Proof System

Recently there has been much academic and industrial interest in practical implementations of zero knowledge proofs. These techniques allow a party to prove to another party that a given statement is true without r

USENIX Security 2018,
Author(s): Howard Wu, Wenting Zheng, Alessandro Chiesa, Raluca Ada Popa, Ion Stoica

Shift: A Zero FLOP, Zero Parameter Alternative to Spatial Convolutions

Neural networks rely on convolutions to aggregate spatial information. However, spatial convolutions are expensive in terms of model size and computation, both of which grow quadratically with respect to kernel siz

Conference on Computer Vision and Pattern Recognition (CVPR),
Author(s): Bichen Wu, Alvin Wan, Xiangyu Yue, Peter Jin, Sicheng Zhao, Noah Golmant, Amir Gholaminejad, Joseph Gonzalez, Kurt Keutzer

NetChain: Scale-Free Sub-RTT Coordination

Coordination services are a fundamental building block of modern cloud systems, providing critical functionalities like configuration management and distributed locking. The major challenge is to achieve low latenc

15th USENIX Symposium on Networked Systems Design and Implementation (NSDI '18), https://www.usenix.org/system/files/conference/nsdi18/nsdi18-jin.pdf
Author(s): Xin Jin, Ion Stoica, Xiaozhou Li, Haoyu Zhang , Nate Foster, Jeongkeun Lee, Robert Soulé, Changhoon Kim

Oblix: An Efficient Oblivious Search Index

Search indices are fundamental building blocks of many systems, and there is great interest in running them on encrypted data. Unfortunately, many known schemes that enable search queries on encrypted data achieve

39th IEEE Symposium on Security and Privacy,
Author(s): Pratyush Mishra, Rishabh Poddar, Jerry Chen, Alessandro Chiesa, Raluca Ada Popa

RLlib: Abstractions for Distributed Reinforcement Learning

Reinforcement learning (RL) algorithms involve the deep nesting of highly irregular computation patterns, each of which typically exhibits opportunities for distributed computation. We argue for distributing

International Conference on Machine Learning (ICML 2018), https://arxiv.org/abs/1712.09381
Author(s): Eric Liang, Richard Liaw, Robert Nishihara, Philipp Moritz, Roy Fox, Kenneth Goldberg, Joseph Gonzalez, Michael Jordan, Ion Stoica

Design and Analysis of a Query Processor for Brick

Brick is a recently proposed metadata schema and ontology for describing building components and the relationships between them. It represents buildings as directed labeled graphs using the RDF data model. Using th

4th ACM International Conference on Systems for Energy-Efficient Built Environments (BuildSys), 2017,
Author(s): Gabe Fierro,

SafeBricks: Shielding Network Functions in the Cloud

With the advent of network function virtualization (NFV), outsourcing network processing to the cloud is growing in popularity amongst enterprises and organizations. Such outsourcing, however, poses a threat to the

15th USENIX Symposium on Networked Systems Design and Implementation (NSDI) 2018, https://rise.cs.berkeley.edu/~rishabhp/publications/SafeBricks.pdf
Author(s): Rishabh Poddar, Chang Lan, Raluca Ada Popa, Sylvia Ratnasamy

Blink: A fast NVLink-based collective communication library

Efficiently training large deep learning models requires scaling training across a number of GPUs. When training at scale, synchronizing parameters across GPUs introduces significant overheads. To improve

SysML 2018, https://rise.cs.berkeley.edu/projects/blink/
Author(s): Guanhua Wang, Amar Phanishayee, Shivaram Venkataraman, Ion Stoica

The emission mobilityscope: personalized data collection for agile urban planning

We have built an initial version of an open source mobilityscope or transportation meter that can track end to end travel patterns using smartphone sensors and user input. The travel patterns can be used to

California Higher Education Sustainability Conference (CHESC), Santa Barbara CA, June 2016, https://chesc.org/wp-content/uploads/shankari_chesc_emission.pdf
Author(s): K. Shankari, Cathy Wu

Optimizing the diamond lane: A more tractable carpool problem and algorithms

Carpooling has been long deemed a promising approach to better utilizing existing transportation infrastructure. However, there are several reasons carpooling is still not the preferred mode of commute in the Unite

19th IEEE Intelligent Transportation Systems Conference, http://ieeexplore.ieee.org/document/7795739/
Author(s): K. Shankari, Ece Kamar, Randy Katz, , Christos Papadimitriou, Eric Horvitz, Alexandre Bayen

Ray: A Distributed Framework for Emerging AI Applications

The next generation of AI applications will continuously interact with the environment and learn from these interactions. These applications impose new and demanding systems requirements, both in terms of performan

USENIX Symposium on Operating Systems Design and Implementation (OSDI), https://arxiv.org/abs/1712.05889
Author(s): Philipp Moritz, Robert Nishihara, Stephanie Wang, Alexey Tumanov, Richard Liaw, Eric Liang, Melih Elibol, Zongheng Yang, William Paul, Michael Jordan, Ion Stoica

DDCO: Discovery of Deep Continuous Options for Robot Learning from Demonstrations

DDCO learns form demonstrations continuous control skills parametrized by deep networks, their termination condition, and high-level policies to use them.

Proceedings of the 1st Annual Conference on Robot Learning (CoRL) 2017, http://proceedings.mlr.press/v78/krishnan17a/krishnan17a.pdf
Author(s): Sanjay Krishnan, Roy Fox, Ion Stoica, Kenneth Goldberg

Sketched Ridge Regression: Optimization Perspective, Statistical Perspective, and Model Averaging

We address the statistical and optimization impacts of using classical sketch versus Hessian sketch to solve approximately the Matrix Ridge Regression (MRR) problem. Prior research has considered the effects of cla

International Conference on Machine Learning, http://proceedings.mlr.press/v70/wang17c/wang17c.pdf
Author(s): Shusen Wang, Alex Gittens, Michael Mahoney

Full-System Simulation of Java Workloads with RISC-V and the Jikes Research Virtual Machine

Managed languages such as Java, JavaScript or Python account for a large portion of workloads, both in cloud data centers and on mobile devices. It is therefore unsurprising that there is an interest in hardware-so

1st Workshop on Computer Architecture Research with RISC-V (CARRV '17), Boston, MA, October 2017, https://people.eecs.berkeley.edu/~maas/papers/maas-carrv17-jikesrvm.pdf
Author(s): Martin Maas, Krste Asanovic, John Kubiatowicz

A Berkeley View of Systems Challenges for AI

In this paper, we propose several open research directions in systems, architectures, and security that can address these challenges and help unlock AI’s potential to improve lives and society.

UC Berkeley EECS Technical Report, https://www2.eecs.berkeley.edu/Pubs/TechRpts/2017/EECS-2017-159.pdf
Author(s): Ion Stoica, Dawn song, Raluca Ada Popa, , Michael Mahoney, Randy Katz, Anthony Joseph, Michael Jordan, Joe Hellerstein, et al.

Finite-Size Corrections and Likelihood Ratio Fluctuations in the Spiked Wigner Model

In this paper we study principal components analysis in the regime of high dimensionality and high noise. Our model of the problem is a rank-one deformation of a Wigner matrix where the signal-to-noise ratio (SNR)

, https://arxiv.org/abs/1710.02903
Author(s): Ahmed El Alaoui, Florent Krzakala, Michael Jordan

Drizzle: Fast and Adaptable Stream Processing at Scale

Large scale streaming systems aim to provide high throughput and low latency. They are often used to run mission-critical applications, and must be available 24×7. Thus such systems need to adapt to failures and in

Symposium on Operating Systems Principles (SOSP) 2017,
Author(s): Shivaram Venkataraman, Aurojit Panda, Kay Ousterhout, Michael Armbrust, Ali Ghodsi, Mike Franklin, Benjamin Recht, Ion Stoica

Capacity Releasing Diffusion for Speed and Locality

Diffusions and related random walk procedures are of central importance in many areas of machine learning, data analysis, and applied mathematics. Because they spread mass agnostically at each step in an iterative

Proceedings of the 34th International Conference on Machine Learning (ICML '17), http://proceedings.mlr.press/v70/wang17b.html
Author(s): Di Wang, Kimon Fountoulakis, Monika Henzinger, Michael W. Mahoney, Satish Rao

A Scalable Distributed Spatial Index for the Internet-of-Things

The increasing interest in the Internet-of-Things (IoT) suggest that a new source of big data is imminent—they would likely be produced by machines and sensors in the IoT ecosystem. The fundamental characteristic

ACM Symposium on Cloud Computing 2017 (SoCC '17),
Author(s): Anand Padmanabha Iyer, Ion Stoica

Learning certifiably optimal rule lists for categorical data

As machine learning continues to gain prominence in socially-important decision-making, the interpretability of predictive models remains a crucial problem. Our goal is to build models that are highly predictive, t

23rd SIGKDD Conference on Knowledge Discovery and Data Mining (KDD '17), https://arxiv.org/abs/1704.01701
Author(s): Elaine Angelino, Nicholas Larus-Stone, Daniel Alabi, Margo Seltzer, Cynthia Rudin

Multi-Task Learning for Straggler Avoiding Predictive Job Scheduling

Parallel processing frameworks (Dean and Ghemawat, 2004) accelerate jobs by breaking them into tasks that execute in parallel. However, slow running or straggler tasks can run up to 8 times slower than the median t

Journal of Machine Learning Research, http://jmlr.org/papers/volume17/15-149/15-149.pdf
Author(s): Neeraja Yadwadkar, Bharath Hariharan, Joseph Gonzalez, Randy Katz

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 accura

ACM Symposium on Cloud Computing 2017 (SoCC '17), https://people.eecs.berkeley.edu/~neerajay/paris_socc17.pdf
Author(s): Neeraja Yadwadkar, Bharath Hariharan, Joseph Gonzalez, Burton Smith, Randy Katz

Data Tweening: Incremental Visualization of Data Transforms

In the context of interactive query sessions, it is common to issue a succession of queries, transforming a dataset to the desired result. It is often difficult to comprehend a succession of transformations, especi

PVLDB, http://www.vldb.org/pvldb/vol10/p661-khan.pdf
Author(s): Larry Xu, Arnab Nandi, Joe Hellerstein

Automating Diagnosis of Cellular Radio Access Network Problems

In an increasingly mobile connected world, our user experience of mobile applications more and more depends on the performance of cellular radio access networks (RANs). To achieve high quality of experience for the

23rd ACM Annual International Conference on Mobile Computing and Networking (MobiCom 2017),
Author(s): Anand Padmanabha Iyer, Li Erran Li, Ion Stoica

Occupy the Cloud: Distributed Computing for the 99%

Distributed computing remains inaccessible to a large number of users, in spite of many open source platforms and extensive commercial offerings. While distributed computation frameworks have moved beyond a simple

ACM Symposium on Cloud Computing (SoCC) 2017, https://arxiv.org/abs/1702.04024
Author(s): Eric Jonas, Qifan Pu, Shivaram Venkataraman, Ion Stoica, Benjamin Recht

Return of the Runtimes: Rethinking the Language Runtime System for the Cloud 3.0 Era

The public cloud is moving to a Platform-as-a-Service model where services such as data management, machine learning or image classification are provided by the cloud operator while applications are written in high

16th Workshop on Hot Topics in Operating Systems (HotOS '17),
Author(s): Martin Maas, Krste Asanovic, John Kubiatowicz

EC-Cache: Load-balanced, Low-latency Cluster Caching with Online Erasure Coding

Data-intensive clusters and object stores are increasingly relying on in-memory object caching to meet the I/O performance demands. These systems routinely face the challenges of popularity skew, background

USENIX OSDI, https://www.usenix.org/conference/osdi16/technical-sessions/presentation/rashmi
Author(s): , Mosharaf Chowdhury, Jack Kosaian, Ion Stoica, Kannan Ramchandran

Breaking Locality Accelerates Block Gauss-Seidel

We analyze the accelerated block Gauss-Seidel algorithm in the random coordinate sampling setting

International Conference on Machine Learning (ICML 2017), https://arxiv.org/abs/1701.03863
Author(s): Shivaram Venkataraman, Stephen Tu, Ashia C. Wilson, Alex Gittens, Michael Jordan, Benjamin Recht

Real-Time Machine Learning: The Missing Pieces

Machine learning applications are increasingly deployed not only to serve predictions using static models, but also as tightly-integrated components of feedback loops involving dynamic, real-time decision

HotOS 2017, https://arxiv.org/abs/1703.03924
Author(s): Robert Nishihara, Philipp Moritz, Stephanie Wang, Alexey Tumanov, William Paul, Johann Schleier-Smith, Richard Liaw, Mehrdad Niknami, Michael Jordan, Ion Stoica

ZipG: A Memory-efficient Graph Store for Interactive Queries

ZipG is a distributed memory-efficient graph store for serving interactive graph queries.

SIGMOD '17, https://people.eecs.berkeley.edu/~anuragk/papers/zipg.pdf
Author(s): Anurag Khandelwal, Zongheng Yang, Evan Ye, Rachit Agarwal, Ion Stoica

Sub-microsecond Adaptive Voltage Scaling in a 28nm FD-SOI Processor

This work presents a RISC-V system-on-chip (SoC) with integrated voltage regulation and power management implemented in 28nm FD-SOI. A fully integrated switched-capacitor DC-DC converter, coupled with an

ESSCIRC - ESSDERC 2016, http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7598294
Author(s): Krste Asanovic, Benjamin Keller, Martin Cochet, Brian Zimmer, Jaehwa Kwak, Alberto Puggelli, Steven Bailey, Borivoje Nikolic, Palmer Dabbelt, et al.

Vector Processors for Energy-Efficient Embedded Systems

High-performance embedded processors are frequently designed as arrays of small, in-order scalar cores, even when their workloads exhibit high degrees of data-level parallelism (DLP). We show that these

ISCA 2016, https://people.eecs.berkeley.edu/~krste/papers/dabbelt-mes2016.pdf
Author(s): Krste Asanovic, Daniel Dabbelt, Colin Schmidt, Eric Love, Howard Mao, Sagar Karandikar

The Renewed Case for the Reduced Instruction Set Computer: Avoiding ISA Bloat with Macro-Op Fusion for RISC-V

This report makes the case that a well-designed Reduced Instruction Set Computer (RISC) can match, and even exceed, the performance and code density of existing commercial Complex Instruction Set Computers (CISC) w

Technical Report, https://people.eecs.berkeley.edu/~krste/papers/EECS-2016-130.pdf
Author(s): Krste Asanovic, Christopher Celio, Daniel Dabbelt, David A. Patterson

Strober: Fast and Accurate Sample-Based Energy Simulation for Arbitrary RTL

This paper presents a sample-based energy simulation methodology that enables fast and accurate estimations of performance and average power for arbitrary RTL designs. Our approach uses an FPGA to simultaneously si

ISCA 2016, https://people.eecs.berkeley.edu/~krste/papers/strober-isca16.pdf
Author(s): Krste Asanovic, Donggyu Kim, Adam Izraelevitz, Christopher Celio, Hokeum Kim, Brian Zimmer

Reprogrammable Redundancy for Cache Vmin Reduction in a 28nm RISC-V Processor

The presented processor lowers SRAM-based cache Vmin by using three architectural techniques–bit bypass (BB), dynamic column redundancy (DCR), and line disable (LD)–that use low-overhead reprogrammable

A-SSCC’16, http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7844150
Author(s): Krste Asanovic, B. Zimmer, P. F. Chiu, Bora Nikolic

High Performance Transactions via Early Write Visibility

In order to guarantee recoverable transaction execution, database systems permit a transaction’s writes to be observable only at the end of its execution. As a consequence, there is generally a delay between the

PVLDB, http://www.vldb.org/pvldb/vol10/p613-faleiro.pdf
Author(s): Joe Hellerstein, Jose M. Faleiro, Daniel J. Abadi

Ground: A Data Context Service

Ground is an open-source data context service , a system to manage all the information that informs the use of data. Data usage has changed both philosophically and practically in the last decade, creating an

CIDR 2017, 8th Biennial Conference on Innovative Data Systems Research, http://cidrdb.org/cidr2017/papers/p111-hellerstein-cidr17.pdf
Author(s): Joe Hellerstein, Joseph Gonzalez, Vikram Sreekanti, et al.

Scalable Atomic Visibility with RAMP Transactions

Databases can provide scalability by partitioning data across several servers. However, multipartition, multioperation transactional access is often expensive, employing coordination-intensive locking, validation,

, https://tinyurl.com/zf5svjr
Author(s): Joe Hellerstein, Ion Stoica, Ali Ghodsi, Alan Fekete, Peter Bailis

Putting logic-based distributed systems on stable grounds

In the Declarative Networking paradigm, Datalog-like languages are used to express distributed computations. Whereas recently formal operational semantics for these languages have been developed, a

Theory and Practice of Logic Programming, Volume 16, Issue 4, https://tinyurl.com/zwb4nf8
Author(s): Joe Hellerstein, TOM J. AMELOOT, JAN VAN DEN BUSSCHE, WILLIAM R. MARCZAK, PETER ALVARO

ReStream: Accelerating Backtesting and Stream Replay with Serial-Equivalent Parallel Processing

Real-time predictive applications can demand continuous and agile development, with new models constantly being trained, tested, and then deployed. Training and testing are done by replaying stored event logs,

SoCC '16 - Proceedings of the Seventh ACM Symposium on Cloud Computing, https://tinyurl.com/h9csy93
Author(s): Johann Schleier-Smith, Erik T. Krogen, Joe Hellerstein

Grail Quest: A New Proposal for Hardware-assisted Garbage Collection

Many big data systems are written in garbage-collected languages and GC has a substantial impact on throughput, responsiveness and predicability of these systems. However, despite decades of research, there is stil

Sixth Workshop on Architectures and Systems for Big Data (ASBD 2016), https://people.eecs.berkeley.edu/~maas/papers/maas-asbd16-hwgc.pdf
Author(s): Martin Maas, Krste Asanovic, John Kubiatowicz

Time-evolving Graph Processing at Scale

Time-evolving graph-structured big data arises naturally in many application domains such as social networks and communication networks. However, existing graph processing systems lack support for efficient computa

Graph Data-management Experiences & Systems (GRADES),
Author(s): Anand Padmanabha Iyer, Li Erran Li, Tathagata Das, Ion Stoica

SparkR: Scaling R Programs with Spark

R is a popular statistical programming language with a number of extensions that support data processing and machine learning tasks. However, interactive data analysis in R is usually limited as the R runtime

,
Author(s): Shivaram Venkataraman, Ali Ghodsi, Ion Stoica

Decentralized Anonymous Micropayments

Anonymity-preserving micropayments for Bitcoin-like currencies.

EUROCRYPT 2017, http://ia.cr/2016/1033
Author(s): Alessandro Chiesa, Matthew Green, Jingcheng Liu, Peihan Miao, Ian Miers, Pratyush Mishra

Iterative methods for solving factorized linear systems

Stochastic iterative algorithms such as the Kaczmarz and Gauss-Seidel methods have gained recent attention because of their speed, simplicity, and the ability to approximately solve large-scale linear systems of eq

, https://arxiv.org/abs/1701.07453
Author(s): Aaditya Ramdas, Anna Ma, Deanna Needell

Generative Models and Model Criticism via Optimized Maximum Mean Discrepancy

We propose a method to optimize the representation and distinguishability of samples from two probability distributions, by maximizing the estimated power of a statistical test based on the maximum mean discrepancy

5th International Conference on Learning Representations, Toulon, 2017, https://arxiv.org/abs/1611.04488
Author(s):

The p-filter: multi-layer FDR control for grouped hypotheses

In many practical applications of multiple hypothesis testing using the False Discovery Rate (FDR), the given hypotheses can be naturally partitioned into groups, and one may not only want to control the number of

Journal of the Royal Statistical Society, Series B (Methodology), https://arxiv.org/abs/1512.03397
Author(s):

MiniCrypt: Reconciling Encryption and Compression for Big Data Stores.

More and more applications and web services generate larger and larger amounts of confidential data, such as user and financial data. On one hand, these systems must use encryption to ensure confidentiality, while

EuroSys 2017, https://pdfs.semanticscholar.org/5aa9/0f7bc1a8282daf74a65aec86022d7ddbad9b.pdf
Author(s): Wenting Zheng, Raluca Ada Popa, Ion Stoica, Rachit Agarwal, Frank Li

Opaque: An Oblivious and Encrypted Distributed Analytics Platform.

As enterprises move to cloud-based analytics, the risk of cloud security breaches poses a serious threat. Encrypting data at rest and in transit is a major first step. However, data must still be decrypted in memor

NSDI 2017 (USENIX Symposium of Networked Systems Design and Implementation),
Author(s): Wenting Zheng, Raluca Ada Popa, Ion Stoica, Joseph Gonzalez, Ankur Dave, Jethro Beekman

Random Projection Design for Scalable Implicit Smoothing of Randomly Observed Stochastic Processes

Standard methods for multi-variate time series analysis are hampered by sampling at random timestamps, long range dependencies , and the scale of the data. In this paper we present a novel estimator for

AISTATS 2017, https://rise.cs.berkeley.edu/wp-content/uploads/2017/03/Aistats_camera_ready.pdf
Author(s): , Joseph Gonzalez, Evan Sparks, Alexandre M. Bayen

Decoding from Pooled data: Phase Transitions of Message Passing

We consider the problem of decoding a discrete signal of categorical variables from the observation of several histograms of pooled subsets of it. We present an Approximate Message Passing (AMP) algorithm for recov

short version submitted to International Symposium on Information Theory (ISIT), long version to be submitted to IEEE Transactions on Information Theory (IEEEIT), https://arxiv.org/abs/1702.02279
Author(s): Aaditya Ramdas, Ahmed El Alaoui, Michael Jordan, Florent Krzakala, Lenka Zdeborova

Decoding from Pooled data: Sharp Information-Theoretic Bounds

Consider a population consisting of n individuals, each of whom has one of d types (e.g. their blood type, in which case d=4). We are allowed to query this database by specifying a subset of the population, and in

submitted to Annals of Applied Probability (AoAP), https://arxiv.org/abs/1611.09981
Author(s): Ahmed El Alaoui, Aaditya Ramdas, Michael Jordan, Florent Krzakala, Lenka Zdeborova

Universality of Mallows’ and degeneracy of Kendall’s kernels for rankings

Kernel methods provide an attractive framework for aggregating and learning from ranking data, and so understanding the fundamental properties of kernels over permutations is a question of broad interest. We provid

, https://arxiv.org/abs/1603.08035
Author(s): Horia Mania, Aaditya Ramdas, Martin J. Wainwright, Michael Jordan, Benjamin Recht

Proteus: agile ML elasticity through tiered reliability in dynamic resource markets

Many shared computing clusters allow users to utilize excess idle resources at lower cost or priority, with the proviso that some or all may be taken away at any time. But, exploiting such dynamic resource availabi

ACM European Conference on Computer Systems (EuroSys'2017), http://www.pdl.cmu.edu/PDL-FTP/BigLearning/proteus_abs.shtml
Author(s): Aaron Harlap, Alexey Tumanov, Andrew Chung, Gregory R. Ganger, Phil Gibbons

Morpheus: Towards Automated SLOs for Enterprise Clusters

Modern resource management frameworks for largescale analytics leave unresolved the problematic tension between high cluster utilization and job’s performance predictability—respectively coveted by operators an

Proceedings of the 12th USENIX Symposium on Operating Systems Design and Implementation (OSDI'16)., https://www.usenix.org/conference/osdi16/technical-sessions/presentation/jyothi
Author(s): C. Curino, I. Menache, S. Narayanamurthy, Alexey Tumanov, J. Yaniv, R. Mavlyutov, I. Goiri, S. Krishnan, J. Kulkarni, S. Rao

Clipper: A Low-Latency Online Prediction Serving System

Machine learning is being deployed in a growing number of applications which demand real-time, accurate, and robust predictions under heavy query load. However, most machine learning frameworks and systems only add

NSDI 2017, https://rise.cs.berkeley.edu/wp-content/uploads/2017/02/clipper_final.pdf
Author(s): Dan Crankshaw, Xin Wang, Giulio Zhou, Michael J. Franklin, Joseph Gonzalez, Ion Stoica

A DeVIL-ish approach to inconsistency in interactive visualizations

Declarative languages have a long tradition in both the database systems and data visualization communities, separating specifications from implementations.Declarative Visual Interaction Language presents a

HILDA 2016, https://nsfdeclarativevis.github.io/NSFDeclarativeVis/devil-hilda16.pdf
Author(s): Yifan Wu, Joe Hellerstein, Eugene Wu