index.json

[{"authors":["admin"],"categories":null,"content":"Chengliang graduated from HKUST under the supervision of Prof. Wei Wang. He was the recipient of the prestigious Hong Kong Ph.D. Fellowship Award (2016 – 2020).\nHis research interests cover the broad area of distributed systems, with a focus on big data and machine learning systems as well as cloud computing. He enjoys identifying fundamental system design and performance issues in large-scale cloud systems and searching for general and efficient solutions.\n","date":1549324800,"expirydate":-62135596800,"kind":"taxonomy","lang":"en","lastmod":1555459200,"objectID":"2525497d367e79493fd32b198b28f040","permalink":"/authors/admin/","publishdate":"0001-01-01T00:00:00Z","relpermalink":"/authors/admin/","section":"authors","summary":"Chengliang graduated from HKUST under the supervision of Prof. Wei Wang. He was the recipient of the prestigious Hong Kong Ph.D. Fellowship Award (2016 – 2020).\nHis research interests cover the broad area of distributed systems, with a focus on big data and machine learning systems as well as cloud computing. He enjoys identifying fundamental system design and performance issues in large-scale cloud systems and searching for general and efficient solutions.","tags":null,"title":"Marc Chengliang Zhang","type":"authors"},{"authors":null,"categories":null,"content":"Flexibility This feature can be used for publishing content such as:\n Online courses Project or software documentation Tutorials  The courses folder may be renamed. For example, we can rename it to docs for software/project documentation or tutorials for creating an online course.\nDelete tutorials To remove these pages, delete the courses folder and see below to delete the associated menu link.\nUpdate site menu After renaming or deleting the courses folder, you may wish to update any [[main]] menu links to it by editing your menu configuration at config/_default/menus.toml.\nFor example, if you delete this folder, you can remove the following from your menu configuration:\n[[main]] name = \u0026#34;Courses\u0026#34; url = \u0026#34;courses/\u0026#34; weight = 50 Or, if you are creating a software documentation site, you can rename the courses folder to docs and update the associated Courses menu configuration to:\n[[main]] name = \u0026#34;Docs\u0026#34; url = \u0026#34;docs/\u0026#34; weight = 50 Update the docs menu If you use the docs layout, note that the name of the menu in the front matter should be in the form [menu.X] where X is the folder name. Hence, if you rename the courses/example/ folder, you should also rename the menu definitions in the front matter of files within courses/example/ from [menu.example] to [menu.\u0026lt;NewFolderName\u0026gt;].\n","date":1536451200,"expirydate":-62135596800,"kind":"section","lang":"en","lastmod":1536451200,"objectID":"59c3ce8e202293146a8a934d37a4070b","permalink":"/courses/example/","publishdate":"2018-09-09T00:00:00Z","relpermalink":"/courses/example/","section":"courses","summary":"Learn how to use Academic's docs layout for publishing online courses, software documentation, and tutorials.","tags":null,"title":"Overview","type":"docs"},{"authors":null,"categories":null,"content":"In this tutorial, I\u0026rsquo;ll share my top 10 tips for getting started with Academic:\nTip 1 Lorem ipsum dolor sit amet, consectetur adipiscing elit. Duis posuere tellus ac convallis placerat. Proin tincidunt magna sed ex sollicitudin condimentum. Sed ac faucibus dolor, scelerisque sollicitudin nisi. Cras purus urna, suscipit quis sapien eu, pulvinar tempor diam. Quisque risus orci, mollis id ante sit amet, gravida egestas nisl. Sed ac tempus magna. Proin in dui enim. Donec condimentum, sem id dapibus fringilla, tellus enim condimentum arcu, nec volutpat est felis vel metus. Vestibulum sit amet erat at nulla eleifend gravida.\nNullam vel molestie justo. Curabitur vitae efficitur leo. In hac habitasse platea dictumst. Sed pulvinar mauris dui, eget varius purus congue ac. Nulla euismod, lorem vel elementum dapibus, nunc justo porta mi, sed tempus est est vel tellus. Nam et enim eleifend, laoreet sem sit amet, elementum sem. Morbi ut leo congue, maximus velit ut, finibus arcu. In et libero cursus, rutrum risus non, molestie leo. Nullam congue quam et volutpat malesuada. Sed risus tortor, pulvinar et dictum nec, sodales non mi. Phasellus lacinia commodo laoreet. Nam mollis, erat in feugiat consectetur, purus eros egestas tellus, in auctor urna odio at nibh. Mauris imperdiet nisi ac magna convallis, at rhoncus ligula cursus.\nCras aliquam rhoncus ipsum, in hendrerit nunc mattis vitae. Duis vitae efficitur metus, ac tempus leo. Cras nec fringilla lacus. Quisque sit amet risus at ipsum pharetra commodo. Sed aliquam mauris at consequat eleifend. Praesent porta, augue sed viverra bibendum, neque ante euismod ante, in vehicula justo lorem ac eros. Suspendisse augue libero, venenatis eget tincidunt ut, malesuada at lorem. Donec vitae bibendum arcu. Aenean maximus nulla non pretium iaculis. Quisque imperdiet, nulla in pulvinar aliquet, velit quam ultrices quam, sit amet fringilla leo sem vel nunc. Mauris in lacinia lacus.\nSuspendisse a tincidunt lacus. Curabitur at urna sagittis, dictum ante sit amet, euismod magna. Sed rutrum massa id tortor commodo, vitae elementum turpis tempus. Lorem ipsum dolor sit amet, consectetur adipiscing elit. Aenean purus turpis, venenatis a ullamcorper nec, tincidunt et massa. Integer posuere quam rutrum arcu vehicula imperdiet. Mauris ullamcorper quam vitae purus congue, quis euismod magna eleifend. Vestibulum semper vel augue eget tincidunt. Fusce eget justo sodales, dapibus odio eu, ultrices lorem. Duis condimentum lorem id eros commodo, in facilisis mauris scelerisque. Morbi sed auctor leo. Nullam volutpat a lacus quis pharetra. Nulla congue rutrum magna a ornare.\nAliquam in turpis accumsan, malesuada nibh ut, hendrerit justo. Cum sociis natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. Quisque sed erat nec justo posuere suscipit. Donec ut efficitur arcu, in malesuada neque. Nunc dignissim nisl massa, id vulputate nunc pretium nec. Quisque eget urna in risus suscipit ultricies. Pellentesque odio odio, tincidunt in eleifend sed, posuere a diam. Nam gravida nisl convallis semper elementum. Morbi vitae felis faucibus, vulputate orci placerat, aliquet nisi. Aliquam erat volutpat. Maecenas sagittis pulvinar purus, sed porta quam laoreet at.\nTip 2 Lorem ipsum dolor sit amet, consectetur adipiscing elit. Duis posuere tellus ac convallis placerat. Proin tincidunt magna sed ex sollicitudin condimentum. Sed ac faucibus dolor, scelerisque sollicitudin nisi. Cras purus urna, suscipit quis sapien eu, pulvinar tempor diam. Quisque risus orci, mollis id ante sit amet, gravida egestas nisl. Sed ac tempus magna. Proin in dui enim. Donec condimentum, sem id dapibus fringilla, tellus enim condimentum arcu, nec volutpat est felis vel metus. Vestibulum sit amet erat at nulla eleifend gravida.\nNullam vel molestie justo. Curabitur vitae efficitur leo. In hac habitasse platea dictumst. Sed pulvinar mauris dui, eget varius purus congue ac. Nulla euismod, lorem vel elementum dapibus, nunc justo porta mi, sed tempus est est vel tellus. Nam et enim eleifend, laoreet sem sit amet, elementum sem. Morbi ut leo congue, maximus velit ut, finibus arcu. In et libero cursus, rutrum risus non, molestie leo. Nullam congue quam et volutpat malesuada. Sed risus tortor, pulvinar et dictum nec, sodales non mi. Phasellus lacinia commodo laoreet. Nam mollis, erat in feugiat consectetur, purus eros egestas tellus, in auctor urna odio at nibh. Mauris imperdiet nisi ac magna convallis, at rhoncus ligula cursus.\nCras aliquam rhoncus ipsum, in hendrerit nunc mattis vitae. Duis vitae efficitur metus, ac tempus leo. Cras nec fringilla lacus. Quisque sit amet risus at ipsum pharetra commodo. Sed aliquam mauris at consequat eleifend. Praesent porta, augue sed viverra bibendum, neque ante euismod ante, in vehicula justo lorem ac eros. Suspendisse augue libero, venenatis eget tincidunt ut, malesuada at lorem. Donec vitae bibendum arcu. Aenean maximus nulla non pretium iaculis. Quisque imperdiet, nulla in pulvinar aliquet, velit quam ultrices quam, sit amet fringilla leo sem vel nunc. Mauris in lacinia lacus.\nSuspendisse a tincidunt lacus. Curabitur at urna sagittis, dictum ante sit amet, euismod magna. Sed rutrum massa id tortor commodo, vitae elementum turpis tempus. Lorem ipsum dolor sit amet, consectetur adipiscing elit. Aenean purus turpis, venenatis a ullamcorper nec, tincidunt et massa. Integer posuere quam rutrum arcu vehicula imperdiet. Mauris ullamcorper quam vitae purus congue, quis euismod magna eleifend. Vestibulum semper vel augue eget tincidunt. Fusce eget justo sodales, dapibus odio eu, ultrices lorem. Duis condimentum lorem id eros commodo, in facilisis mauris scelerisque. Morbi sed auctor leo. Nullam volutpat a lacus quis pharetra. Nulla congue rutrum magna a ornare.\nAliquam in turpis accumsan, malesuada nibh ut, hendrerit justo. Cum sociis natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. Quisque sed erat nec justo posuere suscipit. Donec ut efficitur arcu, in malesuada neque. Nunc dignissim nisl massa, id vulputate nunc pretium nec. Quisque eget urna in risus suscipit ultricies. Pellentesque odio odio, tincidunt in eleifend sed, posuere a diam. Nam gravida nisl convallis semper elementum. Morbi vitae felis faucibus, vulputate orci placerat, aliquet nisi. Aliquam erat volutpat. Maecenas sagittis pulvinar purus, sed porta quam laoreet at.\n","date":1557010800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1557010800,"objectID":"74533bae41439377bd30f645c4677a27","permalink":"/courses/example/example1/","publishdate":"2019-05-05T00:00:00+01:00","relpermalink":"/courses/example/example1/","section":"courses","summary":"In this tutorial, I\u0026rsquo;ll share my top 10 tips for getting started with Academic:\nTip 1 Lorem ipsum dolor sit amet, consectetur adipiscing elit. Duis posuere tellus ac convallis placerat. Proin tincidunt magna sed ex sollicitudin condimentum. Sed ac faucibus dolor, scelerisque sollicitudin nisi. Cras purus urna, suscipit quis sapien eu, pulvinar tempor diam. Quisque risus orci, mollis id ante sit amet, gravida egestas nisl. Sed ac tempus magna. Proin in dui enim.","tags":null,"title":"Example Page 1","type":"docs"},{"authors":null,"categories":null,"content":"Here are some more tips for getting started with Academic:\nTip 3 Lorem ipsum dolor sit amet, consectetur adipiscing elit. Duis posuere tellus ac convallis placerat. Proin tincidunt magna sed ex sollicitudin condimentum. Sed ac faucibus dolor, scelerisque sollicitudin nisi. Cras purus urna, suscipit quis sapien eu, pulvinar tempor diam. Quisque risus orci, mollis id ante sit amet, gravida egestas nisl. Sed ac tempus magna. Proin in dui enim. Donec condimentum, sem id dapibus fringilla, tellus enim condimentum arcu, nec volutpat est felis vel metus. Vestibulum sit amet erat at nulla eleifend gravida.\nNullam vel molestie justo. Curabitur vitae efficitur leo. In hac habitasse platea dictumst. Sed pulvinar mauris dui, eget varius purus congue ac. Nulla euismod, lorem vel elementum dapibus, nunc justo porta mi, sed tempus est est vel tellus. Nam et enim eleifend, laoreet sem sit amet, elementum sem. Morbi ut leo congue, maximus velit ut, finibus arcu. In et libero cursus, rutrum risus non, molestie leo. Nullam congue quam et volutpat malesuada. Sed risus tortor, pulvinar et dictum nec, sodales non mi. Phasellus lacinia commodo laoreet. Nam mollis, erat in feugiat consectetur, purus eros egestas tellus, in auctor urna odio at nibh. Mauris imperdiet nisi ac magna convallis, at rhoncus ligula cursus.\nCras aliquam rhoncus ipsum, in hendrerit nunc mattis vitae. Duis vitae efficitur metus, ac tempus leo. Cras nec fringilla lacus. Quisque sit amet risus at ipsum pharetra commodo. Sed aliquam mauris at consequat eleifend. Praesent porta, augue sed viverra bibendum, neque ante euismod ante, in vehicula justo lorem ac eros. Suspendisse augue libero, venenatis eget tincidunt ut, malesuada at lorem. Donec vitae bibendum arcu. Aenean maximus nulla non pretium iaculis. Quisque imperdiet, nulla in pulvinar aliquet, velit quam ultrices quam, sit amet fringilla leo sem vel nunc. Mauris in lacinia lacus.\nSuspendisse a tincidunt lacus. Curabitur at urna sagittis, dictum ante sit amet, euismod magna. Sed rutrum massa id tortor commodo, vitae elementum turpis tempus. Lorem ipsum dolor sit amet, consectetur adipiscing elit. Aenean purus turpis, venenatis a ullamcorper nec, tincidunt et massa. Integer posuere quam rutrum arcu vehicula imperdiet. Mauris ullamcorper quam vitae purus congue, quis euismod magna eleifend. Vestibulum semper vel augue eget tincidunt. Fusce eget justo sodales, dapibus odio eu, ultrices lorem. Duis condimentum lorem id eros commodo, in facilisis mauris scelerisque. Morbi sed auctor leo. Nullam volutpat a lacus quis pharetra. Nulla congue rutrum magna a ornare.\nAliquam in turpis accumsan, malesuada nibh ut, hendrerit justo. Cum sociis natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. Quisque sed erat nec justo posuere suscipit. Donec ut efficitur arcu, in malesuada neque. Nunc dignissim nisl massa, id vulputate nunc pretium nec. Quisque eget urna in risus suscipit ultricies. Pellentesque odio odio, tincidunt in eleifend sed, posuere a diam. Nam gravida nisl convallis semper elementum. Morbi vitae felis faucibus, vulputate orci placerat, aliquet nisi. Aliquam erat volutpat. Maecenas sagittis pulvinar purus, sed porta quam laoreet at.\nTip 4 Lorem ipsum dolor sit amet, consectetur adipiscing elit. Duis posuere tellus ac convallis placerat. Proin tincidunt magna sed ex sollicitudin condimentum. Sed ac faucibus dolor, scelerisque sollicitudin nisi. Cras purus urna, suscipit quis sapien eu, pulvinar tempor diam. Quisque risus orci, mollis id ante sit amet, gravida egestas nisl. Sed ac tempus magna. Proin in dui enim. Donec condimentum, sem id dapibus fringilla, tellus enim condimentum arcu, nec volutpat est felis vel metus. Vestibulum sit amet erat at nulla eleifend gravida.\nNullam vel molestie justo. Curabitur vitae efficitur leo. In hac habitasse platea dictumst. Sed pulvinar mauris dui, eget varius purus congue ac. Nulla euismod, lorem vel elementum dapibus, nunc justo porta mi, sed tempus est est vel tellus. Nam et enim eleifend, laoreet sem sit amet, elementum sem. Morbi ut leo congue, maximus velit ut, finibus arcu. In et libero cursus, rutrum risus non, molestie leo. Nullam congue quam et volutpat malesuada. Sed risus tortor, pulvinar et dictum nec, sodales non mi. Phasellus lacinia commodo laoreet. Nam mollis, erat in feugiat consectetur, purus eros egestas tellus, in auctor urna odio at nibh. Mauris imperdiet nisi ac magna convallis, at rhoncus ligula cursus.\nCras aliquam rhoncus ipsum, in hendrerit nunc mattis vitae. Duis vitae efficitur metus, ac tempus leo. Cras nec fringilla lacus. Quisque sit amet risus at ipsum pharetra commodo. Sed aliquam mauris at consequat eleifend. Praesent porta, augue sed viverra bibendum, neque ante euismod ante, in vehicula justo lorem ac eros. Suspendisse augue libero, venenatis eget tincidunt ut, malesuada at lorem. Donec vitae bibendum arcu. Aenean maximus nulla non pretium iaculis. Quisque imperdiet, nulla in pulvinar aliquet, velit quam ultrices quam, sit amet fringilla leo sem vel nunc. Mauris in lacinia lacus.\nSuspendisse a tincidunt lacus. Curabitur at urna sagittis, dictum ante sit amet, euismod magna. Sed rutrum massa id tortor commodo, vitae elementum turpis tempus. Lorem ipsum dolor sit amet, consectetur adipiscing elit. Aenean purus turpis, venenatis a ullamcorper nec, tincidunt et massa. Integer posuere quam rutrum arcu vehicula imperdiet. Mauris ullamcorper quam vitae purus congue, quis euismod magna eleifend. Vestibulum semper vel augue eget tincidunt. Fusce eget justo sodales, dapibus odio eu, ultrices lorem. Duis condimentum lorem id eros commodo, in facilisis mauris scelerisque. Morbi sed auctor leo. Nullam volutpat a lacus quis pharetra. Nulla congue rutrum magna a ornare.\nAliquam in turpis accumsan, malesuada nibh ut, hendrerit justo. Cum sociis natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. Quisque sed erat nec justo posuere suscipit. Donec ut efficitur arcu, in malesuada neque. Nunc dignissim nisl massa, id vulputate nunc pretium nec. Quisque eget urna in risus suscipit ultricies. Pellentesque odio odio, tincidunt in eleifend sed, posuere a diam. Nam gravida nisl convallis semper elementum. Morbi vitae felis faucibus, vulputate orci placerat, aliquet nisi. Aliquam erat volutpat. Maecenas sagittis pulvinar purus, sed porta quam laoreet at.\n","date":1557010800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1557010800,"objectID":"1c2b5a11257c768c90d5050637d77d6a","permalink":"/courses/example/example2/","publishdate":"2019-05-05T00:00:00+01:00","relpermalink":"/courses/example/example2/","section":"courses","summary":"Here are some more tips for getting started with Academic:\nTip 3 Lorem ipsum dolor sit amet, consectetur adipiscing elit. Duis posuere tellus ac convallis placerat. Proin tincidunt magna sed ex sollicitudin condimentum. Sed ac faucibus dolor, scelerisque sollicitudin nisi. Cras purus urna, suscipit quis sapien eu, pulvinar tempor diam. Quisque risus orci, mollis id ante sit amet, gravida egestas nisl. Sed ac tempus magna. Proin in dui enim. Donec condimentum, sem id dapibus fringilla, tellus enim condimentum arcu, nec volutpat est felis vel metus.","tags":null,"title":"Example Page 2","type":"docs"},{"authors":["Chengliang Zhang","Junzhe Xia","Baichen Yang","Huangcheng Puyang","Wei Wang","Ruichuan Chen","Istemi Ekin Akkus","Paarijaat Aditya","Feng Yan"],"categories":null,"content":"","date":1629936000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1629936000,"objectID":"9c008dc05a7a67acf32223da82722302","permalink":"/publication/citadel/","publishdate":"2021-08-24T00:00:00Z","relpermalink":"/publication/citadel/","section":"publication","summary":"To be updated.","tags":["Machine Learning","Data Privacy","Intel SGX"],"title":"Citadel: Protecting Data Privacy and Model Confidentiality for Collaborative Learning","type":"publication"},{"authors":["Chengliang Zhang"],"categories":null,"content":"","date":1629936000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1629936000,"objectID":"0f40cf545adf79432be77d9037b7794b","permalink":"/publication/thesis/","publishdate":"2021-08-24T00:00:00Z","relpermalink":"/publication/thesis/","section":"publication","summary":"Please refer to the enclosed pdf.","tags":["Machine Learning","Data Privacy","Intel SGX"],"title":"Towards Efficient and Secure Large-Scale Systems for Distributed Machine Learning Training","type":"publication"},{"authors":["Chengliang Zhang","Suyi Li","Junzhe Xia","Wei Wang","Feng Yan","Yang Liu"],"categories":null,"content":"","date":1594512000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1594512000,"objectID":"4e2a59315ff7eddd05cafc322ddd39af","permalink":"/publication/batchcrypt/","publishdate":"2020-04-24T00:00:00Z","relpermalink":"/publication/batchcrypt/","section":"publication","summary":"Cross-silo federated learning (FL) enables organizations (e.g., financial or medical) to collaboratively train a machine learning model by aggregating local gradient updates from each client without sharing privacy-sensitive data. To ensure no update is revealed during aggregation, industrial FL frameworks allow clients to mask local gradient updates using additively homomorphic encryption (HE). However, this results in significant cost in computation and communication. In our characterization, HE operations dominate the training time, while inflating the data transfer amount by two orders of magnitude. In this paper, we present BatchCrypt, a system solution for cross-silo FL that substantially reduces the encryption and communication overhead caused by HE. Instead of encrypting individual gradients with full precision, we encode a batch of quantized gradients into a long integer and encrypt it in one go. To allow gradient-wise aggregation to be performed on ciphertexts of the encoded batches, we develop new quantization and encoding schemes along with a novel gradient clipping technique. We implemented BatchCrypt as a plugin module in FATE, an industrial cross-silo FL framework. Evaluations with EC2 clients in geo-distributed datacenters show that BatchCrypt achieves 23X-93X training speedup while reducing the communication overhead by 66X-101X. The accuracy loss due to quantization errors is less than 1%.","tags":["Machine Learning","Federated learning"],"title":"BatchCrypt: Efficient Homomorphic Encryption for Cross-Silo Federated Learning","type":"publication"},{"authors":["Chengliang Zhang","Minchen Yu","Wei Wang","Feng Yan"],"categories":null,"content":"","date":1593475200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1593475200,"objectID":"0d4a120f7645de863ecdcbe1cf500a75","permalink":"/publication/mark-journal/","publishdate":"2020-04-24T00:00:00Z","relpermalink":"/publication/mark-journal/","section":"publication","summary":"The remarkable advances of Machine Learning (ML) have spurred an increasing demand for ML-as-a-Service on public cloud - developers train and publish ML models as online services to provide low-latency inference for dynamic queries. The primary challenge of ML model serving is to meet the response-time Service-Level Objectives (SLOs) of inference workloads while minimizing serving cost. In this paper, we proposes MArk (Model Ark), a general-purpose inference serving system, to tackle the dual challenge of SLO compliance and cost effectiveness. MArk employs three design choices tailored to inference workload. First, MArk dynamically batches requests and opportunistically serves them using expensive hardware accelerators (e.g., GPU) for improved performance-cost ratio. Second, instead of relying on feedback control scaling or over-provisioning to serve dynamic workload, which can be too slow or too expensive, MArk employs predictive autoscaling to hide the provisioning latency at low cost. Third, given the stateless nature of inference serving, MArk exploits the flexible, yet costly serverless instances to cover occasional load spikes that are hard to predict. We evaluated the performance of MArk using several state-of-the-art ML models trained in TensorFlow, MXNet, and Keras. Compared with the premier industrial ML serving platform SageMaker, MArk reduces the serving cost up to 7.8 tiimes while achieving even better latency performance.","tags":["Machine Learning"],"title":"Enabling Cost-Effective, SLO-Aware Machine Learning Inference Serving on Public Cloud","type":"publication"},{"authors":["Jun Yi","Chengliang Zhang","Wei Wang","Cheng Li","Feng Yan"],"categories":null,"content":"","date":1589241600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1589241600,"objectID":"2ce9208582c82da0da85e42943f7425c","permalink":"/publication/heterbo/","publishdate":"2019-12-07T00:00:00Z","relpermalink":"/publication/heterbo/","section":"publication","summary":"Machine-Learning-as-a-Service (MLaaS) enables practitioners and AI service providers to train and deploy ML models in the cloud using diverse and scalable compute resources. A common problem for MLaaS users is to choose from a variety of training deployment options, notably scale-up (using more capable instances) and scale-out (using more instances), subject to the budget limits and/or time constraints. State-of-theart (SOTA) approaches employ analytical modeling for finding the optimal deployment strategy. However, they have limited applicability as they must be tailored to specific ML model architectures, training framework, and hardware. To quickly adapt to the fast evolving design of ML models and hardware infrastructure, we propose a new Bayesian Optimization (BO) based method HeterBO for exploring the optimal deployment of training jobs. Unlike the existing BO approaches for general applications, we consider the heterogeneous exploration cost and machine learning specific prior to significantly improve the search efficiency. This paper culminates in a fully automated MLaaS training Cloud Deployment system (MLCD) driven by the highly efficient HeterBO search method. We have extensively evaluated MLCD in AWS EC2, and the experimental results show that MLCD outperforms two SOTA baselines, conventional BO and CherryPick, by 3.1X and 2.34X, respectively.","tags":["Machine Learning"],"title":"Not All Explorations Are Equal: Harnessing Heterogeneous Profiling Cost for Efficient MLaaS Training","type":"publication"},{"authors":["Yinghao Yu","Chengliang Zhang","Wei Wang","Jun Zhang","Khaled Ben Letaief"],"categories":null,"content":"","date":1567296000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1567296000,"objectID":"deef6f1413718ce7e02fdc53c7365895","permalink":"/publication/lrc/","publishdate":"2019-09-01T00:00:00Z","relpermalink":"/publication/lrc/","section":"publication","summary":"Memory caches are being used aggressively in today’s data analytics systems such as Spark, Tez, and Piccolo. The significant performance impact of caches and their limited sizes call for efficient cache management in data analytics clusters. However, prevalent data analytics systems employ rather simple cache management policies—notably Least Recently Used (LRU) and Least Frequently Used (LFU)—that are oblivious to the application semantics of data dependency, expressed as directed acyclic graphs (DAGs). Without this knowledge, cache management can, at best, be performed by “guessing” the future data access patterns based on history, which frequently results in inefficient, erroneous caching with a low hit rate and a long response time. Worse still, the lack of data dependency knowledge makes it impossible to retain the all-or-nothing cache property of cluster applications, in that a compute task cannot be sped up unless all the dependent data has been kept in the main memory. In this paper, we propose a novel cache replacement policy, named Least Reference Count (LRC), which exploits the application’s data dependency information to optimize the cache management. LRC keeps track of the reference count of each data block, defined as the number of dependent child blocks that have not been computed yet, and always evicts the block with the smallest reference count. Furthermore, we incorporate the all-or-nothing requirement into LRC by coordinately managing the reference counts of all the input data blocks for the same computation. We demonstrate the efficacy of LRC through both empirical analysis and cluster deployments against popular benchmarking workloads. Our Spark implementation shows that, the proposed policies well address the all-or-nothing requirement and significantly improve the cache performance. Compared with LRU and a recently proposed caching policy called MEMTUNE, LRC improves the caching performance of typical workloads in production clusters by 22% and 284%, respectively.","tags":["Spark"],"title":"Towards Dependency-Aware Cache Management for Data Analytics Applications","type":"publication"},{"authors":["Chengliang Zhang","Minchen Yu","Wei Wang","Feng Yan"],"categories":null,"content":"","date":1562889600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1562889600,"objectID":"f242ac020155eab3ee1d74b2e0244623","permalink":"/publication/mark/","publishdate":"2017-01-01T00:00:00Z","relpermalink":"/publication/mark/","section":"publication","summary":"The advances of Machine Learning (ML) have sparked a growing demand of ML-as-a-Service, developers train ML models and publish them in the cloud as online services to provide low-latency inference at scale. The key challenge of ML model serving is to meet the response-time Service-Level-Objectives (SLOs) of inference workloads while minimizing the serving cost. In this paper, we tackle the dual challenge of SLO compliance and cost effectiveness with MArk (Model Ark), a general-purpose inference serving system built in Amazon Web Services (AWS). MArk employs three design choices tailor-made for inference workload. First, MArk dynamically batches requests and opportunistically serves them using expensive hardware accelerators (e.g., GPU) for improved performance-cost ratio. Second, instead of relying on feedback control scaling or over-provisioning to serve dynamic workload, which can be too slow or too expensive for inference serving, MArk employs predictive autoscaling to hide the provisioning latency at low cost. Third, given the stateless nature of inference serving, MArk exploits the flexible, yet costly serverless instances to cover the occasional load spikes that are hard to predict. We evaluated the performance of MArk using several state-of-the-art ML models trained in popular frameworks including TensorFlow, MXNet, and Keras. Compared with the premier industrial ML serving platform SageMaker, MArk reduces the serving cost up to 7.8x while achieving even better latency performance.","tags":["Machine Learning"],"title":"MArk: Exploiting Cloud Services for Cost-Effective, SLO-Aware Machine Learning Inference Serving","type":"publication"},{"authors":["Marc Chengliang Zhang"],"categories":[],"content":"from IPython.core.display import Image Image(\u0026#39;https://www.python.org/static/community_logos/python-logo-master-v3-TM-flattened.png\u0026#39;) print(\u0026#34;Welcome to Academic!\u0026#34;) Welcome to Academic!  Install Python and Jupyter Install Anaconda which includes Python 3 and Jupyter notebook.\nOtherwise, for advanced users, install Jupyter notebook with pip3 install jupyter.\nCreate a new blog post as usual Run the following commands in your Terminal, substituting \u0026lt;MY_WEBSITE_FOLDER\u0026gt; and my-post with the file path to your Academic website folder and a name for your blog post (without spaces), respectively:\ncd \u0026lt;MY_WEBSITE_FOLDER\u0026gt; hugo new --kind post post/my-post cd \u0026lt;MY_WEBSITE_FOLDER\u0026gt;/content/post/my-post/ Create or upload a Jupyter notebook Run the following command to start Jupyter within your new blog post folder. Then create a new Jupyter notebook (New \u0026gt; Python Notebook) or upload a notebook.\njupyter notebook Convert notebook to Markdown jupyter nbconvert Untitled.ipynb --to markdown --NbConvertApp.output_files_dir=. # Copy the contents of Untitled.md and append it to index.md: cat Untitled.md | tee -a index.md # Remove the temporary file: rm Untitled.md Edit your post metadata Open index.md in your text editor and edit the title etc. in the front matter according to your preference.\nTo set a featured image, place an image named featured into your post\u0026rsquo;s folder.\nFor other tips, such as using math, see the guide on writing content with Academic.\n","date":1549324800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1549324800,"objectID":"6e929dc84ed3ef80467b02e64cd2ed64","permalink":"/post/jupyter/","publishdate":"2019-02-05T00:00:00Z","relpermalink":"/post/jupyter/","section":"post","summary":"Learn how to blog in Academic using Jupyter notebooks","tags":[],"title":"Display Jupyter Notebooks with Academic","type":"post"},{"authors":["Chengliang Zhang","Minchen Yu","Wei Wang","Feng Yan"],"categories":null,"content":"","date":1531353600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1531353600,"objectID":"28b56e622340bdbd30843489ec006e85","permalink":"/publication/spec-sync/","publishdate":"2018-01-01T00:00:00Z","relpermalink":"/publication/spec-sync/","section":"publication","summary":"Large machine learning models are typically trained in parallel and distributed environments. The model parameters are iteratively refined by multiple worker nodes in parallel, each processing a subset of the training data. In practice, the training is usually conducted in an asynchronous parallel manner, where workers can proceed to the next iteration before receiving the latest model parameters. While this maximizes the rate of updates, the price paid is compromised training quality as the computation is usually performed using stale model parameters. To address this problem, we propose a new scheme, termed speculative synchronization. Our scheme allows workers to speculate about the recent parameter updates from others on the fly, and if necessary, the workers abort the ongoing computation, pull fresher parameters, and start over to improve the quality of training. We design an effective heuristic algorithm to judiciously determine when to restart training iterations with fresher parameters by quantifying the gain and loss. We implement our scheme in MXNet—a popular machine learning framework—and demonstrate its effectiveness through cluster deployment atop Amazon EC2. Experimental results show that speculative synchronization achieves up to 3x speedup over the asynchronous parallel scheme in many machine learning applications, with little additional communication overhead.","tags":["Machine Learning"],"title":"Stay Fresh: Speculative Synchronization for Fast Distributed Machine Learning","type":"publication"},{"authors":null,"categories":null,"content":"","date":1461715200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1461715200,"objectID":"071bf75b2ad9bc1e099bec284e9c20b2","permalink":"/project/spark-cache-management/","publishdate":"2016-04-27T00:00:00Z","relpermalink":"/project/spark-cache-management/","section":"project","summary":"Participated in the implementation of online LRC module in Spark.","tags":["Spark"],"title":"LRC: Dependency-Aware Cache Management for Data Analytics Clusters","type":"project"},{"authors":["Marc Chengliang Zhang"],"categories":[],"content":"Create a free website with Academic using Markdown, Jupyter, or RStudio. Choose a beautiful color theme and build anything with the Page Builder - over 40 widgets, themes, and language packs included!\nCheck out the latest demo of what you\u0026rsquo;ll get in less than 10 minutes, or view the showcase of personal, project, and business sites.\n Setup Academic Get Started View the documentation Ask a question Request a feature or report a bug Updating? View the Update Guide and Release Notes Support development of Academic:  Donate a coffee Become a backer on Patreon Decorate your laptop or journal with an Academic sticker Wear the T-shirt    \nKey features:\n Page builder - Create anything with widgets and elements Edit any type of content - Blog posts, publications, talks, slides, projects, and more! Create content in Markdown, Jupyter, or RStudio Plugin System - Fully customizable color and font themes Display Code and Math - Code highlighting and LaTeX math supported Integrations - Google Analytics, Disqus commenting, Maps, Contact Forms, and more! Beautiful Site - Simple and refreshing one page design Industry-Leading SEO - Help get your website found on search engines and social media Media Galleries - Display your images and videos with captions in a customizable gallery Mobile Friendly - Look amazing on every screen with a mobile friendly version of your site Multi-language - 15+ language packs including English, 中文, and Português Multi-user - Each author gets their own profile page Privacy Pack - Assists with GDPR Stand Out - Bring your site to life with animation, parallax backgrounds, and scroll effects One-Click Deployment - No servers. No databases. Only files.  Color Themes Academic comes with day (light) and night (dark) mode built-in. Click the sun/moon icon in the top right of the Demo to see it in action!\nChoose a stunning color and font theme for your site. Themes are fully customizable and include:\n         Ecosystem  Academic Admin: An admin tool to import publications from BibTeX or import assets for an offline site Academic Scripts: Scripts to help migrate content to new versions of Academic  Install You can choose from one of the following four methods to install:\n one-click install using your web browser (recommended) install on your computer using Git with the Command Prompt/Terminal app install on your computer by downloading the ZIP files install on your computer with RStudio  Then personalize and deploy your new site.\nUpdating View the Update Guide.\nFeel free to star the project on Github to help keep track of updates.\nLicense Copyright 2016-present George Cushen.\nReleased under the MIT license.\n","date":1461110400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1555459200,"objectID":"279b9966ca9cf3121ce924dca452bb1c","permalink":"/post/getting-started/","publishdate":"2016-04-20T00:00:00Z","relpermalink":"/post/getting-started/","section":"post","summary":"Create a beautifully simple website in under 10 minutes.","tags":["Academic"],"title":"Academic: the website builder for Hugo","type":"post"},{"authors":null,"categories":null,"content":"","date":1430092800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1430092800,"objectID":"8450badc71362979dae86f30e31027c6","permalink":"/project/du-robot/","publishdate":"2015-04-27T00:00:00Z","relpermalink":"/project/du-robot/","section":"project","summary":"Participated in the research and development of a series of smart robots, as a engineer of the DeepQA module.","tags":["Demo"],"title":"DuRobot","type":"project"}]