arXiv subjects

Minjie Wang

Publications and source records attributed to Minjie Wang.

41 records · Page 3Linked to original sources

arXiv2019

Deterministic generation and partial retrieval of a spin-wave excitation in an atomic ensemble

In this paper, we report a generation of a spin-wave excitation (SWE) with a near-unity () probability in a given time (~730). Such deterministic generation relies on a feedback scheme with a millisecond quantum memory. The millisecond memory is achieved by maximizing the wavelength of the spin wave and storing the SWE as the magnetic-field-insensitive transition. We then demonstrate partial retrievals of the spin wave by applying a first read pulse whose area is smaller than the value of . The remained SWE is fully retrieved by a second pulse. Anti-correlation function between the detections in the first and second readouts has been measured, which shows that the partial-retrieval operation on the SWE is in the quantum regime. The presented experiment represents an important step towards the realization of the improved DLCZ quantum repeater protocol proposed in Phys. Rev. A 77, 062301 (2008).

quant-ph↗

arXiv2019

Supporting Very Large Models using Automatic Dataflow Graph Partitioning

This paper presents Tofu, a system that partitions very large DNN models across multiple GPU devices to reduce per-GPU memory footprint. Tofu is designed to partition a dataflow graph of fine-grained tensor operators in order to work transparently with a general-purpose deep learning platform like MXNet. In order to automatically partition each operator, we propose to describe the semantics of an operator in a simple language which represents tensors as lambda functions mapping from tensor coordinates to values. To optimally partition different operators in a dataflow graph, Tofu uses a recursive search algorithm that minimizes the total communication cost. Our experiments on an 8-GPU machine show that Tofu enables the training of very large CNN and RNN models. It also achieves 25% - 400% speedup over alternative approaches to train very large models.

cs.DC↗

arXiv2019

Learned Indexes for Dynamic Workloads

The recent proposal of learned index structures opens up a new perspective on how traditional range indexes can be optimized. However, the current learned indexes assume the data distribution is relatively static and the access pattern is uniform, while real-world scenarios consist of skew query distribution and evolving data. In this paper, we demonstrate that the missing consideration of access patterns and dynamic data distribution notably hinders the applicability of learned indexes. To this end, we propose solutions for learned indexes for dynamic workloads (called Doraemon). To improve the latency for skew queries, Doraemon augments the training data with access frequencies. To address the slow model re-training when data distribution shifts, Doraemon caches the previously-trained models and incrementally fine-tunes them for similar access patterns and data distribution. Our preliminary result shows that, Doraemon improves the query latency by 45.1% and reduces the model re-training time to 1/20.

cs.LG↗

arXiv2018

Unifying Data, Model and Hybrid Parallelism in Deep Learning via Tensor Tiling

Deep learning systems have become vital tools across many fields, but the increasing model sizes mean that training must be accelerated to maintain such systems' utility. Current systems like Tensorflow and MXNet focus on one specific parallelization strategy, data parallelism, which requires large training batch sizes in order to scale. We cast the problem of finding the best parallelization strategy as the problem of finding the best tiling to partition tensors with the least overall communication. We propose an algorithm that can find the optimal tiling. Our resulting parallelization solution is a hybrid of data parallelism and model parallelism. We build the SoyBean system that performs automatic parallelization. SoyBean automatically transforms a serial dataflow graph captured by an existing deep learning system frontend into a parallel dataflow graph based on the optimal tiling it has found. Our evaluations show that SoyBean is 1.5x-4x faster than pure data parallelism for AlexNet and VGG. We present this automatic tiling in a new system, SoyBean, that can act as a backend for Tensorflow, MXNet, and others.

cs.DC↗

arXiv2015

MXNet: A Flexible and Efficient Machine Learning Library for Heterogeneous Distributed Systems

MXNet is a multi-language machine learning (ML) library to ease the development of ML algorithms, especially for deep neural networks. Embedded in the host language, it blends declarative symbolic expression with imperative tensor computation. It offers auto differentiation to derive gradients. MXNet is computation and memory efficient and runs on various heterogeneous systems, ranging from mobile devices to distributed GPU clusters. This paper describes both the API design and the system implementation of MXNet, and explains how embedding of both symbolic expression and tensor operation is handled in a unified fashion. Our preliminary experiments reveal promising results on large scale deep neural network applications using multiple GPU machines.

cs.DC↗