Filters








10,444 Hits in 7.0 sec

Deep Reinforcement Learning for Join Order Enumeration

Ryan Marcus, Olga Papaemmanouil
2018 Proceedings of the First International Workshop on Exploiting Artificial Intelligence Techniques for Data Management - aiDM'18  
In this paper, we argue that existing deep reinforcement learning techniques can be applied to address this challenge.  ...  Join order selection plays a significant role in query performance.  ...  RL-Based Join Enumeration Here, we present a proof-of-concept deep reinforcement learning join order enumerator, which we call ReJOIN.  ... 
doi:10.1145/3211954.3211957 dblp:conf/sigmod/MarcusP18 fatcat:jlu6usqdi5f25dvd7mpowkxezy

Towards a Hands-Free Query Optimizer through Deep Learning [article]

Ryan Marcus, Olga Papaemmanouil
2018 arXiv   pre-print
We identify potential complications for future research that integrates deep learning with query optimization, and we describe three novel deep learning based approaches that can lead the way to end-to-end  ...  In this vision paper, we argue that a new type of query optimizer, based on deep reinforcement learning, can drastically improve on the state-of-the-art.  ...  Counter-intuitively, ReJOIN's deep reinforcement learning algorithm (after training) is faster than Post-greSQL's built-in join order enumerator in many cases.  ... 
arXiv:1809.10212v2 fatcat:k4srbxjwy5fd7nd7nqdcwpnoci

Learning to Optimize Join Queries With Deep Reinforcement Learning [article]

Sanjay Krishnan, Zongheng Yang, Ken Goldberg, Joseph Hellerstein, Ion Stoica
2019 arXiv   pre-print
Recognizing the link between classical Dynamic Programming enumeration methods and recent results in Reinforcement Learning (RL), we propose a new method for learning optimized join search strategies.  ...  Exhaustive enumeration of all possible join orders is often avoided, and most optimizers leverage heuristics to prune the search space.  ...  Our key insight is that join ordering has a deep algorithmic connection with Reinforcement Learning (RL) [47] . Join ordering's sequential structure is the same problem structure that underpins RL.  ... 
arXiv:1808.03196v2 fatcat:w35hsr6li5g23lbhnw6z3ueibe

Join Query Optimization with Deep Reinforcement Learning Algorithms [article]

Jonas Heitz, Kurt Stockinger
2019 arXiv   pre-print
Recent success of deep reinforcement learning (DRL) creates new opportunities for the field of query optimization to tackle the above-mentioned problems.  ...  The main idea of FOOP is to use a data-adaptive learning query optimizer that avoids exhaustive enumerations of join orders and is thus significantly faster than traditional approaches based on dynamic  ...  We also want to thank Katharina Rombach for the many discussions about reinforcement learning.  ... 
arXiv:1911.11689v1 fatcat:oxdertk5frfynhmdnnt7td3hz4

Learning State Representations for Query Optimization with Deep Reinforcement Learning [article]

Jennifer Ortiz, Magdalena Balazinska, Johannes Gehrke, S. Sathiya Keerthi
2018 arXiv   pre-print
Deep reinforcement learning is quickly changing the field of artificial intelligence.  ...  Our goal in this work is to explore the capabilities of deep reinforcement learning in the context of query optimization.  ...  [11] uses a deep reinforcement learning technique to determine join order for a fixed database.  ... 
arXiv:1803.08604v1 fatcat:h2cytvx55re5hfzkdrg2l4gfoq

Simpli-Squared: A Very Simple Yet Unexpectedly Powerful Join Ordering Algorithm Without Cardinality Estimates [article]

Asoke Datta, Yesdaulet Izenov, Brian Tsan, Florin Rusu
2021 arXiv   pre-print
The join order of a given query is computed by splitting the join graph along the many-to-many joins and sorting the tables based on their size.  ...  Simpli-Squared computes the join order without using any statistics or cardinality estimates.  ...  For join-ordering, reinforcement learning is of particular interest. Krishnan et al.  ... 
arXiv:2111.00163v1 fatcat:u7wsj2xx6jau7dxgxgqljfcapa

Similarity Modeling on Heterogeneous Networks via Automatic Path Discovery [article]

Carl Yang, Mengxiong Liu, Frank He, Xikun Zhang, Jian Peng, Jiawei Han
2019 arXiv   pre-print
To this end, we combine continuous reinforcement learning and deep content embedding into a novel semi-supervised joint learning framework.  ...  Specifically, the supervised reinforcement learning component explores useful paths between a small set of example similar pairs of nodes, while the unsupervised deep embedding component captures node  ...  For AutoPath, we train the reinforcement learning agent with the training data and deep embedding on the whole network.  ... 
arXiv:1910.01448v1 fatcat:vmpwqapycvd7xk4iyjmiawq4eq

Neo: A Learned Query Optimizer [article]

Ryan Marcus, Parimarjan Negi, Hongzi Mao, Chi Zhang, Mohammad Alizadeh, Tim Kraska, Olga Papaemmanouil, Nesime Tatbul
2019 arXiv   pre-print
relies on deep neural networks to generate query executions plans.  ...  Despite the progress made over the past decades, query optimizers remain extremely complex components that require a great deal of hand-tuning for specific workloads and datasets.  ...  ReJOIN [34] proposed a deep reinforcement learning approach for join order enumeration [34] , which was generalized into a broader vision for designing an end-to-end learningbased query optimizer in  ... 
arXiv:1904.03711v1 fatcat:sm5bmuavtvffrjkg5qno7pwaaq

Neo

Ryan Marcus, Parimarjan Negi, Hongzi Mao, Chi Zhang, Mohammad Alizadeh, Tim Kraska, Olga Papaemmanouil, Nesime Tatbul
2019 Proceedings of the VLDB Endowment  
relies on deep neural networks to generate query executions plans.  ...  Despite the progress made over the past decades, query optimizers remain extremely complex components that require a great deal of hand-tuning for specific workloads and datasets.  ...  Neo iteratively improves its performance through a combination of reinforcement learning and a search strategy.  ... 
doi:10.14778/3342263.3342644 fatcat:ep3tixgsbrdxhlt4ekm2flvc6u

Solving System Problems with Machine Learning

Ion STOICA
2019 Studies in Informatics and Control  
Over the past decade, Machine Learning (ML) has achieved tremendous successes and has seen wide-scale adoption for human-facing tasks, such as visual recognition, speech recognition, language translation  ...  However, going forward, we contend that ML has an even higher potential for impact by solving hard systems problems, such as improving industrial processes, supply chain optimization, and automatic program  ...  An RL algorithm that uses a DNN to approximate the policy is referred to as a Deep RL algorithm. Figure 1 shows the components of the reinforcement learning system.  ... 
doi:10.24846/v28i2y201901 fatcat:fktydwhmzzgvrexswwe246k4pm

Learned Cardinalities: Estimating Correlated Joins with Deep Learning [article]

Andreas Kipf, Thomas Kipf, Bernhard Radke, Viktor Leis, Peter Boncz, Alfons Kemper
2018 arXiv   pre-print
We describe a new deep learning approach to cardinality estimation.  ...  Our evaluation of MSCN using a real-world dataset shows that deep learning significantly enhances the quality of cardinality estimation, which is the core problem in query optimization.  ...  RELATED WORK Deep learning has been applied to query optimization by three recent papers [13, 23, 27 ] that formulate join ordering as a reinforcement learning problem and use ML to find query plans.  ... 
arXiv:1809.00677v2 fatcat:2oqbfpvop5h2pbsovvha5p3xzq

A Survey on Deep Reinforcement Learning for Data Processing and Analytics [article]

Qingpeng Cai, Can Cui, Yiyuan Xiong, Wei Wang, Zhongle Xie, Meihui Zhang
2021 arXiv   pre-print
Recently, reinforcement learning, deep reinforcement learning (DRL) in particular, is increasingly explored and exploited in many areas because it can learn better strategies in complicated environments  ...  First, we present an introduction to key concepts, theories, and methods in deep reinforcement learning.  ...  Notably, deep reinforcement learning (DRL), which incorporates deep learning (DL) techniques to handle complex unstructured data, has been designed to learn from historical data and self exploration to  ... 
arXiv:2108.04526v2 fatcat:civvkztyx5br7gdp62stt25r7y

A Unified Transferable Model for ML-Enhanced DBMS [article]

Ziniu Wu, Pei Yu, Peilun Yang, Rong Zhu, Yuxing Han, Yaliang Li, Defu Lian, Kai Zeng, Jingren Zhou
2021 arXiv   pre-print
Recently, the database management system (DBMS) community has witnessed the power of machine learning (ML) solutions for DBMS tasks.  ...  Moreover, for each retraining, they require an excessive amount of training data, which is very expensive to acquire and unavailable for a new DB.  ...  For clarity of discussion, we focus on generating the left-deep join orders [17] , which can be directly flattened into an ordered sequence of tables.  ... 
arXiv:2105.02418v3 fatcat:ljb66dxlkvhdtbwnam73e5mxm4

How I Learned to Stop Worrying and Love Re-optimization [article]

Matthew Perron, Zeyuan Shang, Tim Kraska, Michael Stonebraker
2019 arXiv   pre-print
We demonstrate that re-optimization improves the end-to-end latency of the top 20 longest running queries in the Join Order Benchmark by 27%, realizing most of the benefit of perfect cardinality estimation  ...  Cost-based query optimizers remain one of the most important components of database management systems for analytic workloads.  ...  Cuttlefish, for example cannot change the join order, but chooses operator implementations at runtime with a reinforcement learning approach [9] .  ... 
arXiv:1902.08291v2 fatcat:kodfhpdykneejbmx5trjrqy5km

QR-SDN: Towards Reinforcement Learning States, Actions, and Rewards for Direct Flow Routing in Software-Defined Networks

Justus Rischke, Peter Sossalla, Hani Salah, Frank H. P. Fitzek, Martin Reisslein
2020 IEEE Access  
We also acknowledge that this study focuses on reinforcement learning and does not consider deep reinforcement learning.  ...  , and rewards for reinforcement learning.  ... 
doi:10.1109/access.2020.3025432 fatcat:myaoxypjazbcjoohw6u5c7t4ne
« Previous Showing results 1 — 15 out of 10,444 results