Search for: All records

With increasingly deployed cameras and the rapid advances of Computer Vision, large-scale live video analytics becomes feasible. However, analyzing videos is compute-intensive. In addition, live video analytics needs to be performed in real time. In this paper, we design an edge server system for live video analytics. We propose to perform configuration adaptation without profiling video online. We select configurations with a prediction model based on object movement features. In addition, we reduce the latency through resource orchestration on video analytics servers. The key idea of resource orchestration is to batch inference tasks that use the same CNN model, and schedule tasks based on a priority value that estimates their impact on the total latency. We evaluate our system with two video analytic applications, road traffic monitoring and pose detection. The experimental results show that our profiling-free adaptation reduces the workload by 80% of the state-of-the-art adaptation without lowering the accuracy. The average serving latency is reduced by up to 95% comparing with the profiling-based adaptation. 

Graph convolutional network (GCN) has been shown effective in many applications with graph structures. However, training a large-scale GCN is still challenging due to the high computation cost that grows with the size of the graph. In this paper, we propose CM-GCN, a distributed GCN framework using cohesive mini-batches to accelerate large-scale GCN training. The cohesive mini-batches group nodes that are tightly connected in the graph. As a result, CM-GCN can reduce the computation required to train a GCN. We propose a computation cost function to quantify the computation required for mini-batches. By exploring the submodular property of the computation cost function, we develop an efficient algorithm to partition nodes into tightly coupled mini-batches. Based on the computation cost function, we evenly distribute the workloads of mini-batches to workers. We design asynchronous computations between GCN layers to further eliminating the waiting among workers. We implement a CM-GCN framework and evaluate its performance with graphs that contain millions of nodes. Our evaluation shows that CM-GCN can achieve up to 3X speedup without compromising the training accuracy. 

Data parallel frameworks become essential for training machine learning models. The classic Bulk Synchronous Parallel (BSP) model updates the model parameters through pre-defined synchronization barriers. However, when a worker computes significantly slower than other workers, waiting for the slow worker will lead to excessive waste of computing resources. In this paper, we propose a novel proactive data-parallel (PDP) framework. PDP enables the parameter server to initiate the update of the model parameter. That is, we can perform the update at any time without pre-defined update points. PDP not only initiates the update but also determines when to update. The global decision on the frequency of updates will accelerate the training. We further propose asynchronous PDP to reduce the idle time caused by synchronizing parameter updates. We theoretically prove the convergence property of asynchronous PDP. We implement a distributed PDP framework and evaluate PDP with several popular machine learning algorithms including Multilayer Perceptron, Convolutional Neural Network, K-means, and Gaussian Mixture Model. Our evaluation shows that PDP can achieve up to 20X speedup over the BSP model and scale to large clusters. 

Machine learning models such as deep neural networks have been shown to be successful in solving a wide range of problems. Training such a model typically requires stochastic gradient descent, and the process is time-consuming and expensive in terms of computing resources. In this paper, we propose a distributed framework that supports the prioritized execution of the gradient computation. Our proposed distributed framework identifies important data points through computing or estimating the priority for each data point. We evaluate the proposed distributed framework with several machine learning models including multi-layer perceptron (MLP) and convolutional neural networks (CNN). Our experimental results show that prioritized SGD accelerates the training of machine learning models by as much as 1.6X over that of the mini-batch SGD. Further, the distributed framework scales linearly with the number of workers. 

Video analytics has many applications in traffic control, security monitoring, action/event analysis, etc. With the adoption of deep neural networks, the accuracy of video analytics in video streams has been greatly improved. However, deep neural networks for performing video analytics are compute-intensive. In order to reduce processing time, many systems switch to the lower frame rate or resolution. State-of-the-art switching approaches adjust configurations by profiling video clips on a large configuration space. Multiple configurations are tested periodically and the cheapest one with a desired accuracy is adopted. In this paper, we propose a method that adapts the configuration by analyzing past video analytics results instead of profiling candidate configurations. Our method adopts a lower/higher resolution or frame rate when objects move slow/fast. We train a model that automatically selects the best configuration. We evaluate our method with two real-world video analytics applications: traffic tracking and pose estimation. Compared to the periodic profiling method, our method achieves 3%-12% higher accuracy with the same resource cost and 8-17x faster with comparable accuracy. 

Cloud platforms often execute parallel batch applications, such as distributed machine learning (ML), that include numerous synchronization barriers. These barriers, which prevent any task from advancing beyond a specified point until all tasks have reached that point, significantly degrade application performance by reducing it to that of the slowest "straggler" task. To address the problem, researchers have proposed numerous straggler mitigation techniques, including speculatively re-executing straggler tasks and various relaxations of strict barrier semantics. While these techniques improve parallel application performance, they incur a cost in terms of the resources wasted re-executing tasks or waiting. Importantly, these costs, which are often implicit in prior work that targets dedicated resources, become explicit in the cloud, which charges for resources at fine-grained intervals. In addition, the cost difference between techniques is exacerbated in cloud platforms, since they charge substantially less for transient resources that effectively yield a probabilistic performance across a wide range. While transient resources' low list price is attractive, revocations increase the frequency and severity of stragglers, which decreases parallel job performance and increases overall execution cost. To better understand the cost of synchronization, we develop simple analytical models of different straggler mitigation techniques and compare their cost and performance on on-demand and transient resources. Our analysis shows that i) transient servers offer complex tradeoffs compared to on-demand servers, and can result in higher overall costs despite their highly discounted price due to their probabilistic performance; ii) common approaches to straggler mitigation, which is a well-studied problem, are less effective using transient servers that cause frequent and severe stragglers; and iii) a recent approach to flexible synchronization offers the best cost and performance.