Elastic Bulk Synchronous Parallel Model for Distributed Deep Learning

arXiv.org Machine Learning

--The bulk synchronous parallel (BSP) is a celebrated synchronization model for general-purpose parallel computing that has successfully been employed for distributed training of machine learning models. A prevalent shortcoming of the BSP is that it requires workers to wait for the straggler at every iteration. BSP a model that aims to relax its strict synchronization requirement. The proposed model offers more flexibility and adaptability during the training phase, without sacrificing on the accuracy of the trained model. It also achieves comparable (if not higher) accuracy than the other sensible synchronization models. The parameter server framework [1] [2] has been widely adopted to distributing the training of large deep neural network (DNN) models [3] [4]. The framework consists of multiple workers and a logical server that maintains globally shared parameters, typically represented as dense or sparse vectors and matrices [5], and it supports two approaches: model parallelism and data parallelism [6]. In this paper we focus on data parallelism. Data parallelism refers to partitioning (sharding) of large training data into smaller equal size shards and assigning them to workers. Then, the entire DNN model is replicated to each worker.

Normalized Spectral Map Synchronization

Neural Information Processing Systems

The algorithmic advancement of synchronizing maps is important in order to solve a wide range of practice problems with possible large-scale dataset. In this paper, we provide theoretical justifications for spectral techniques for the map synchronization problem, i.e., it takes as input a collection of objects and noisy maps estimated between pairs of objects, and outputs clean maps between all pairs of objects. We show that a simple normalized spectral method that projects the blocks of the top eigenvectors of a data matrix to the map space leads to surprisingly good results. As the noise is modelled naturally as random permutation matrix, this algorithm NormSpecSync leads to competing theoretical guarantees as state-of-the-art convex optimization techniques, yet it is much more efficient. We demonstrate the usefulness of our algorithm in a couple of applications, where it is optimal in both complexity and exactness among existing methods.

Cooperative Learning of Audio and Video Models from Self-Supervised Synchronization

Neural Information Processing Systems

There is a natural correlation between the visual and auditive elements of a video. In this work we leverage this connection to learn general and effective models for both audio and video analysis from self-supervised temporal synchronization. We demonstrate that a calibrated curriculum learning scheme, a careful choice of negative examples, and the use of a contrastive loss are critical ingredients to obtain powerful multi-sensory representations from models optimized to discern temporal synchronization of audio-video pairs. Without further fine-tuning, the resulting audio features achieve performance superior or comparable to the state-of-the-art on established audio classification benchmarks (DCASE2014 and ESC-50). At the same time, our visual subnet provides a very effective initialization to improve the accuracy of video-based action recognition models: compared to learning from scratch, our self-supervised pretraining yields a remarkable gain of 19.9% in action recognition accuracy on UCF101 and a boost of 17.7% on HMDB51.

BML: A High-performance, Low-cost Gradient Synchronization Algorithm for DML Training

Neural Information Processing Systems

In distributed machine learning (DML), the network performance between machines significantly impacts the speed of iterative training. In this paper we propose BML, a new gradient synchronization algorithm with higher network performance and lower network cost than the current practice. BML runs on BCube network, instead of using the traditional Fat-Tree topology. BML algorithm is designed in such a way that, compared to the parameter server (PS) algorithm on a Fat-Tree network connecting the same number of server machines, BML achieves theoretically 1/k of the gradient synchronization time, with k/5 of switches (the typical number of k is 2∼4). Experiments of LeNet-5 and VGG-19 benchmarks on a testbed with 9 dual-GPU servers show that, BML reduces the job completion time of DML training by up to 56.4%.

Translation Synchronization via Truncated Least Squares

Neural Information Processing Systems

In this paper, we introduce a robust algorithm, \textsl{TranSync}, for the 1D translation synchronization problem, in which the aim is to recover the global coordinates of a set of nodes from noisy measurements of relative coordinates along an observation graph. The basic idea of TranSync is to apply truncated least squares, where the solution at each step is used to gradually prune out noisy measurements. We analyze TranSync under both deterministic and randomized noisy models, demonstrating its robustness and stability. Experimental results on synthetic and real datasets show that TranSync is superior to state-of-the-art convex formulations in terms of both efficiency and accuracy. Papers published at the Neural Information Processing Systems Conference.