ProjecToR: Agile Reconfigurable Data Center Interconnect
- Monia Ghobadi, Ratul Mahajan, D. Kilper
- Computer ScienceConference on Applications, Technologies…
- 22 August 2016
Simulations using realistic data center workloads show that this novel, free-space optics based approach for building data center interconnects can improve mean flow completion time by 30-95% and reduce cost by 25-40%.
Collecting Telemetry Data Privately
- Bolin Ding, Janardhan Kulkarni, S. Yekhanin
- Computer ScienceNIPS
- 4 December 2017
This paper develops new LDP mechanisms geared towards repeated collection of counter data, with formal privacy guarantees even after being executed for an arbitrarily long period of time, which have been deployed by Microsoft to collect telemetry across millions of devices.
GRAPHENE: Packing and Dependency-Aware Scheduling for Data-Parallel Clusters
- Robert Grandl, Srikanth Kandula, Sriram Rao, Aditya Akella, Janardhan Kulkarni
- Business, Computer ScienceUSENIX Symposium on Operating Systems Design and…
- 8 August 2016
A newcluster scheduler aimed at jobs that have a complex dependency structure and heterogeneous resource demands, which can compute a DAG schedule, offline, by first scheduling such troublesome tasks and then scheduling the remaining tasks without violating dependencies.
Morpheus: Towards Automated SLOs for Enterprise Clusters
- S. Jyothi, C. Curino, Sriram Rao
- Computer ScienceUSENIX Symposium on Operating Systems Design and…
- 2 November 2016
Morpheus is a new system that codifies implicit user expectations as explicit Service Level Objectives (SLOs) inferred from historical data, enforces SLOs using novel scheduling techniques that isolate jobs from sharing-induced performance variability, and mitigates inherent performance variance by means of dynamic reprovisioning of jobs.
Differentially Private Fine-tuning of Language Models
- Da Yu, Saurabh Naik, Huishuai Zhang
- Computer ScienceInternational Conference on Learning…
- 13 October 2021
All the experiments suggest that larger models are better suited for private fine-tuning: while they are well known to achieve superior accuracy non-privately, they find that they also better maintain their accuracy when privacy is introduced.
Robust Price of Anarchy Bounds via LP and Fenchel Duality
- Janardhan Kulkarni, V. Mirrokni
- EconomicsACM-SIAM Symposium on Discrete Algorithms
- 4 January 2015
A new framework based on LP and Fenchel duality for bounding the robust price of anarchy for a large class of games is developed and used to give the first PoA bounds for temporal routing games on graphs and energy minimization games in machine scheduling.
Algorithms for Cost-Aware Scheduling
- Janardhan Kulkarni, Kamesh Munagala
- Computer ScienceWorkshop on Approximation and Online Algorithms
- 13 September 2012
In this paper, we generalize classical machine scheduling problems by introducing a cost involved in processing jobs, which varies as a function of time. Before defining the problems formally and…
Differentially Private Set Union
- Sivakanth Gopi, P. Gulhane, Janardhan Kulkarni, J. Shen, Milad Shokouhi, S. Yekhanin
- Computer ScienceInternational Conference on Machine Learning
- 22 February 2020
Two new algorithms for differentially private set union are designed, one using Laplace noise and other Gaussian noise, which use $\ell_1$- contractive and $\ell-2$-contractive policies respectively and provide concrete examples of such policies.
Private Non-smooth Empirical Risk Minimization and Stochastic Convex Optimization in Subquadratic Steps
- Janardhan Kulkarni, Y. Lee, Daogao Liu
- Computer ScienceArXiv
- 29 March 2021
This work gets a (nearly) optimal bound on the excess empirical risk and excess population loss with subquadratic gradient complexity on the differentially private Empirical Risk Minimization and Stochastic Convex Optimization problems for non-smooth convex functions.
SelfishMigrate: A Scalable Algorithm for Non-clairvoyantly Scheduling Heterogeneous Processors
- Sungjin Im, Janardhan Kulkarni, Kamesh Munagala, K. Pruhs
- Computer ScienceIEEE Annual Symposium on Foundations of Computer…
- 7 April 2014
This work presents the first online algorithm that is scalable ((1 + ϵ)-speed O(1/2)-competitive for any constant ϵ > 0) for the total weighted flow-time objective and demonstrates the usefulness of ideas from coordination mechanisms and Nash equilibria for designing and analyzing online algorithms.