Filters








3,314 Hits in 2.2 sec

An Imitation Learning Approach for Cache Replacement [article]

Evan Zheran Liu, Milad Hashemi, Kevin Swersky, Parthasarathy Ranganathan, Junwhan Ahn
2020 arXiv   pre-print
In contrast, we propose an imitation learning approach to automatically learn cache access patterns by leveraging Belady's, an oracle policy that computes the optimal eviction decision given the future  ...  As a result, current replacement policies typically resort to heuristics designed for specific common access patterns, which fail on more diverse and complex access patterns.  ...  Acknowledgements We thank Zhan Shi for insightful discussions and for providing the results for Glider, which we compare against.  ... 
arXiv:2006.16239v2 fatcat:v2gm2hjanrdhnafysbswk7dsku

Machine Learning-Based Cache Replacement Policies: A Survey

Pratheeksha P, Student, Department of Computer Science, RV College of Engineering, Bangalore, India., Revathi S A, Assistant Professor, Department of Computer Science, RV College of Engineering, Bangalore, India.
2021 International Journal of Engineering and Advanced Technology  
Despite extensive developments in improving cache hit rates, designing an optimal cache replacement policy that mimics Belady's algorithm still remains a challenging task.  ...  In this paper, we review some of the machine-learning based cache replacement policies that outperformed the static heuristics.  ...  PARROT PARROT is the first method to construct an end -to-end cache replacement policy using imitation learning that approximates Belady's [20] .  ... 
doi:10.35940/ijeat.f2907.0810621 fatcat:jfkkubfxtvgy3fw4uenbdagvpq

SmartChoices: Hybridizing Programming and Machine Learning [article]

Victor Carbune, Thierry Coppey, Alexander Daryin, Thomas Deselaers, Nikhil Sarda, Jay Yagnik
2019 arXiv   pre-print
Our implementation relies on standard Reinforcement Learning (RL) methods. To learn faster, we use the heuristic function, which they are replacing, as an initial function.  ...  We present SmartChoices, an approach to making machine learning (ML) a first class citizen in programming languages which we see as one way to lower the entrance cost to applying ML to problems in new  ...  Acknowledgements The authors are part of a larger effort aiming to hybridize machine learning and programming.  ... 
arXiv:1810.00619v3 fatcat:5wmwihxf4za73nnp7bl7gpy6xe

DEAP Cache: Deep Eviction Admission and Prefetching for Cache [article]

Ayush Mangal, Jitesh Jain, Keerat Kaur Guliani, Omkar Bhalerao
2020 arXiv   pre-print
Recent approaches for learning policies to improve caching, target just one out of the prefetching, admission and eviction processes.  ...  We present our approach as a "proof of concept" of learning all three components of cache strategies using machine learning and leave improving practical deployment for future work.  ...  [3] used an imitation learning-based approach for cache replacement, wherein they used a byte-level representation to deal with the exponential size of address vocabulary.  ... 
arXiv:2009.09206v1 fatcat:o7nke3srwbgaplikcqt3vhkwjy

Phoebe: Reuse-Aware Online Caching with Reinforcement Learning for Emerging Storage Models [article]

Nan Wu, Pengcheng Li
2020 arXiv   pre-print
To this end, we propose Phoebe, a reuse-aware reinforcement learning framework for the optimal online caching that is applicable for a wide range of emerging storage models.  ...  Experiment results show that Phoebe is able to close the gap of cache miss rate from LRU and a state-of-the-art online learning based cache policy to the Belady's optimal policy by 70.3% and 52.6%, respectively  ...  It requires to run OPT on a predefined trace in advance so that it can imitate behaviors of OPT, and thus it is not an online approach. Configurations.  ... 
arXiv:2011.07160v1 fatcat:lzpz6jrw6zgo7gdz4sybod6qzi

Analyzing a Caching Model [article]

Leon Sixt, Evan Zheran Liu, Marie Pellat, James Wexler, Milad Hashemi, Been Kim, Martin Maas
2022 arXiv   pre-print
By analyzing a state-of-the-art caching model, we provide evidence that the model has learned concepts beyond simple statistics that can be leveraged for explanations.  ...  Machine Learning has been successfully applied in systems applications such as memory prefetching and caching, where learned models have been shown to outperform heuristics.  ...  The general strategy of these approaches is to replace a hard-coded system policy, such as a caching policy or a branch predictor, with a learned model.  ... 
arXiv:2112.06989v2 fatcat:x3b6qrzqfvhlpki3gfcqtdoasa

A Review on Adaptive Web Caching Technique

Pranay Nanda, Shamsher Singh, G. L.
2016 International Journal of Computer Applications  
Primary cache replacement algorithms consider arrival time as the only one factor as the basis of their functionality.  ...  The response times of Web accesses are in the order of seconds (versus milliseconds for system access), which allows for more elaborate caching algorithms.  ...  Shamsher Singh for his most support and encouragement. He kindly read my paper and offered invaluable detailed advices on grammar, organization, and the theme of the paper.  ... 
doi:10.5120/ijca2016908043 fatcat:jakazw7ifrhnbbst5vfb2zepqi

Imitation of Success Leads to Cost of Living Mediated Fairness in the Ultimatum Game [article]

Yunong Chen and Andrew Belmonte and Christopher Griffin
2020 arXiv   pre-print
In this paper, we analyze a social imitation model that incorporates internal energy caches (e.g., food/money savings), cost of living, death, and reproduction.  ...  We show that when imitation (and death) occurs, a natural correlation between selfishness and cost of living emerges.  ...  Rajtmajer for her feedback on earlier drafts. CG thanks R. Bailey (USN) who ran the very first simulation of this phenomena in MATLAB while at the United States Naval Academy.  ... 
arXiv:2009.01970v2 fatcat:45kxejmgtbdsnjeywgrmw4hyom

A Taxonomy of ML for Systems Problems

Martin Maas
2020 IEEE Micro  
different approaches for applying machine learning in systems.  ...  We describe a taxonomy to help identify whether or not machine learning should be applied to particular systems problems, and which approaches are most promising.  ...  Figure 2 . 2 ML for memory allocation. 8 (a) Visualizing memory fragmentation. (b) Reinforcement learning approach. (c) Imitation learning approach.  ... 
doi:10.1109/mm.2020.3012883 fatcat:2f43mmsvnvdj5co5dem4ptcvhu

RC-RNN: Reconfigurable Cache Architecture for Storage Systems Using Recurrent Neural Networks

Shahriar Ebrahimi, Reza Salkhordeh, Seyed Ali Osia, Ali Taheri, Hamid R. Rabiee, Hossein Asadi
2021 IEEE Transactions on Emerging Topics in Computing  
In this paper, we propose RC-RNN, the first reconfigurable SSD-based cache architecture for storage systems that utilizes machine learning to identify performance-critical data pages for I/O caching.  ...  Experimental results show that RC-RNN characterizes workloads with an accuracy up to 94.6% for SNIA I/O workloads.  ...  However, such machine learning approaches require few hundreds of microseconds for processing.  ... 
doi:10.1109/tetc.2021.3102041 fatcat:gyqnb3526fbebohwr4qapeiche

A Symmetric Multiprocessor Architecture for Multi-Agent Temporal Difference Learning

Scott Fields, Itamar Elhanany
2006 2006 49th IEEE International Midwest Symposium on Circuits and Systems  
To the author's knowledge, this is the first application of an SMP architecture to a multi-agent reinforcement learning system.  ...  In addition to correctness, one metric of a technique's performance is its learning rate -the number of iterations required to converge to an optimal solution.  ...  The authors would like to thank Zhenzhen Liu for her many insightful suggestions and comments regarding this work. This work has been partially supported by the Woodrow W. Everett, Jr.  ... 
doi:10.1109/mwscas.2006.382109 fatcat:il226hmlyfejdmjv7closzqkuy

Learning from Routing Information for Detecting Routing Misbehavior in Ad Hoc Networks

Robert Basomingera, Young-June Choi
2020 Sensors  
Despite different heuristic approaches that have been proposed, there is still room for improvement.  ...  The detection system learns from shared routing information and uses supervised learning, when previous network status or an exploratory network is available, to train the model, or it uses unsupervised  ...  For each algorithm, parameters were calibrated by a grid search approach.  ... 
doi:10.3390/s20216275 pmid:33158143 fatcat:nv6sd4yihndzpdugykzxjdha6m

A Deep Reinforcement Learning-Based Caching Strategy for IoT Networks with Transient Data [article]

Hongda Wu, Ali Nasehzadeh, Ping Wang
2022 arXiv   pre-print
Adopting deep reinforcement learning (DRL) algorithms enables us to develop an effective caching scheme without the need for any prior knowledge or contextual information.  ...  An efficient caching policy can help meet the standard quality of service requirements while bypassing IoT networks' specific limitations.  ...  To obtain an optimal policy through value function or Q-function is known as value-based reinforcement learning approach.  ... 
arXiv:2203.12674v1 fatcat:zjyrhuntarfprbwprlqg5y5xeu

Social Cognition and the Evolution of Language: Constructing Cognitive Phylogenies

W. Tecumseh Fitch, Ludwig Huber, Thomas Bugnyar
2010 Neuron  
The major and minor clades help to contextualize the phylogenetic position of these species utilizing traditional Linnaean classifications, even when (as for class "Reptilia") this traditional grouping  ...  ACKNOWLEDGMENTS We thank Nadja Kavcik for preparing the figures, and two anonymous reviewers for their helpful comments on an earlier version of the paper.  ...  An often implicit assumption is that living in social groups favors the evolution of social learning, leading to the idea that social learning is an adaptation for social living.  ... 
doi:10.1016/j.neuron.2010.03.011 pmid:20346756 pmcid:PMC4415479 fatcat:ubcxmtxf65cf7m2rjh4cqskjw4

Hybrid Policy Learning for Energy-Latency Tradeoff in MEC-Assisted VR Video Service [article]

Chong Zheng and Shengheng Liu and Yongming Huang and Luxi Yang
2021 arXiv   pre-print
After jointly assessing the caching and computing capacities on both the MEC server and the VR playback device, a hybrid policy is then implemented to coordinate the dynamic caching replacement and the  ...  The underlying multi-objective problem is reformulated as a partially observable Markov decision process, and a deep deterministic policy gradient algorithm is proposed to iteratively learn its solution  ...  In this paper we focus on the deep reinforcement learning approach for the formulated problem (21) .  ... 
arXiv:2104.01036v1 fatcat:gsuujyravfcxdainllrz35fjgy
« Previous Showing results 1 — 15 out of 3,314 results