Filters








116,492 Hits in 3.0 sec

Memory Exploitation in Learning Classifier Systems

Robert E. Smith
1994 Evolutionary Computation  
Learning classifier systems, genetic algorithms, reinforcement learning, bucket brigade algorithm, memory, parasites.  ...  Final comments present future directions for research on memory exploitation in the LCS and similar evolutionary computing systems.  ...  Learning classifier systems (LCSs) (Holland, Holyoak, Nisbett, & Thagard, 1986 ) offer a unique opportunity to examine adaptive memory exploitation.  ... 
doi:10.1162/evco.1994.2.3.199 fatcat:4lvocumhubc4xaxxotbnskj6e4

A metric for software vulnerabilities classification [article]

Gabriele Modena
2014 arXiv   pre-print
Vulnerability discovery and exploits detection are two wide areas of study in software engineering.  ...  A relation between the classifier choice and the features has also been outlined.  ...  may result in injection into memory and execution of arbitrary code that may result in an attacker exploiting the system.  ... 
arXiv:1212.3669v2 fatcat:pcke667kbbhqxkxogebu43kh6m

An autonomous explore/exploit strategy

Alex McMahon, Dan Scott, Will Browne
2005 Proceedings of the 2005 workshops on Genetic and evolutionary computation - GECCO '05  
The XCS learning classifier system uses a fixed explore/exploit balance, but does keep multiple statistics about its performance and interaction in an environment.  ...  In reinforcement learning problems it has been considered that neither exploitation nor exploration can be pursued exclusively without failing at the task.  ...  AI ARCHITECTURES Learning Classifier Systems Learning Classifier Systems are rule-based evolutionary learning systems.  ... 
doi:10.1145/1102256.1102280 dblp:conf/gecco/McMahonSB05 fatcat:vuqkmebcmfgz3cqdhwq73rtkci

Summoning Demons: The Pursuit of Exploitable Bugs in Machine Learning [article]

Rock Stevens, Octavian Suciu, Andrew Ruef, Sanghyun Hong, Michael Hicks, Tudor Dumitraş
2017 arXiv   pre-print
While prior work in the field of adversarial machine learning has studied the impact of input manipulation on correct ML algorithms, we consider the exploitation of bugs in ML implementations.  ...  We propose a semi-automated technique, called steered fuzzing, for exploring this attack surface and for discovering exploitable bugs in machine learning programs, in order to demonstrate the magnitude  ...  However, the adversary may achieve her goals with less powerful logical exploits, e.g., targeting memory corruption bugs that allow modifying data in memory but do not enable code execution or bugs that  ... 
arXiv:1701.04739v1 fatcat:nwkbcoh2ozhy7pmtmtxv4jctti

Analysis of Linux Kernel Vulnerabilities

Supriya Raheja, Geetika Munjal, Shagun ,
2016 Indian Journal of Science and Technology  
Application: Accurately classifying integrity will help in making the system more secure and vulnerable proof.  ...  The performance of these classifiers is evaluated on the basis of accuracy. Findings: All classifiers are showing highest accuracy in terms of integrity.  ...  Classifiers Classification is a type of supervised learning in which a labeled training data is provided from which some output is concluded 15 .  ... 
doi:10.17485/ijst/2016/v9i48/105819 fatcat:acott3wke5aojoma2ulgkrw2b4

Tracking Recurrent Concepts Using Context [chapter]

João Bártolo Gomes, Ernestina Menasalvas, Pedro A. C. Sousa
2010 Lecture Notes in Computer Science  
Moreover, the memory costs associated with the approach are analyzed and the proposed memory-aware strategy is tested, showing that despite the memory consumption cost the learning process accuracy increases  ...  The approach combines on the one hand the performance of stored models representing previously learned concepts and on the other hand exploits learned relations between context and stored models.  ...  Experiments to analyze the performance of the learning system in situations of memory scarcity have been conducted.  ... 
doi:10.1007/978-3-642-13529-3_19 fatcat:u3honnr7qffnpdig2jkla6ozli

Class-incremental Learning with Pre-allocated Fixed Classifiers [article]

Federico Pernici, Matteo Bruni, Claudio Baecchi, Francesco Turchini, Alberto Del Bimbo
2020 arXiv   pre-print
To address this problem, effective methods exploit past data stored in an episodic memory while expanding the final classifier nodes to accommodate the new classes.  ...  In class-incremental learning, a learning agent faces a stream of data with the goal of learning new classes while not forgetting previous ones.  ...  Dual-memory incremental learning is exploited in [36] to keep track of statistics of past classes, in order to rebalance their prediction scores in later stages of learning.  ... 
arXiv:2010.08657v1 fatcat:ul553pejwbauxob2xbpb6g7eau

Self-Learning Pipeline for Low-Energy Resource-Constrained Devices

Fouad Sakr, Riccardo Berta, Joseph Doyle, Alessandro De De Gloria, Francesco Bellotti
2021 Energies  
This paper introduces the self-learning autonomous edge learning and inferencing pipeline (AEP), deployable in a resource-constrained embedded system, which can be used for unsupervised local training  ...  The paper makes an in-depth performance analysis of the system, particularly addressing the limited memory footprint of embedded devices and the need to support remote training robustness.  ...  We have explored feasibility of a self-supervised learning system in an embedded environment in [7] .  ... 
doi:10.3390/en14206636 fatcat:6hdfi7mebbax5ldk2mivcwlhya

Page 205 of Computational Linguistics Vol. 27, Issue 2 [page]

2001 Computational Linguistics  
Memory-based learning is a learning method that is based on storing all examples of a task in memory and then classifying new examples by similarity-based reasoning from these stored examples.  ...  These types of extended second-level features can be exploited by WPDYV, as well as by a wide selection of other machine learning algorithms. 2.3 Memory-based Combination Our first choice from these other  ... 

Smaller Is Better: An Analysis of Instance Quantity/Quality Trade-off in Rehearsal-based Continual Learning [article]

Francesco Pelosin, Andrea Torsello
2022 arXiv   pre-print
to learning systems.  ...  Within this context, rehearsal-based methods i.e., solutions in where the learner exploits memory to revisit past data, has proven to be very effective, leading to performance at the state-of-the-art.  ...  300KiB to 600KiB when then the same systems adopt complex classifiers using several megabytes of memory just for the learned parameters and in the order of gigabytes of working memory for learning.  ... 
arXiv:2105.14106v4 fatcat:qckaifggjnftjmaayezephhn2y

Predicting GPU Performance from CPU Runs Using Machine Learning

Ioana Baldini, Stephen J. Fink, Erik Altman
2014 2014 IEEE 26th International Symposium on Computer Architecture and High Performance Computing  
two out Benchmarks • Parboil 2.0 • Rodinia 2.2 System • Dual-processor Intel Xeon X5690 • 12 cores total @ 3.47 GHz • ATI FirePro v9800 • Nvidia Tesla C2050 Classifier Accuracy SVM NNGE  ...  SIMD Memory Memory loads LD Memory stores ST Memory fences FENCE Control Flow Conditional and unconditional branches BR OpenMP Speedup of 12 threads over sequential execution OMP  ...  Learning the speedup factor of GPU execution (SVM) 5 Binary classifiers: "GPU Speedup > 1" "GPU Speedup > 2" "GPU Speedup > 3" "GPU Speedup > 4" "GPU Speedup > 5" Classifier Accuracy Summary of Results  ... 
doi:10.1109/sbac-pad.2014.30 dblp:conf/sbac-pad/BaldiniFA14 fatcat:h4lwrcnlzvhbpcgssn2ll24n3a

Accelerating Random Forest training process using FPGA

Chuan Cheng, Christos-Savvas Bouganis
2013 2013 23rd International Conference on Field programmable Logic and Applications  
enable full exploitation of the high memory bandwidth offered by the on-chip memory featured on FPGA devices.  ...  Random Forest (RF) is one of the state-of-art supervised learning methods in Machine Learning and inherently consists of two steps: the training and the evaluation step.  ...  ACCELERATE RF TRAINING ON FPGA Batch learning Training a RF involves intense memory access making the access to the off-chip memory the bottleneck of an FPGAbased system.  ... 
doi:10.1109/fpl.2013.6645500 dblp:conf/fpl/ChengB13 fatcat:qhwhtjln6rbwniyomejhb5syry

Low Power and Scalable Many-Core Architecture for Big-Data Stream Computing

Karim Kanoun, Martino Ruggiero, David Atienza, Mihaela van der Schaar
2014 2014 IEEE Computer Society Annual Symposium on VLSI  
small-scale computing infrastructures or current embedded systems), or they are simply dedicated to a specific learning algorithm (i.e., limited to run with a single type of classifiers).  ...  In order to cope with the Big-Data stream input and its high variability, modern stream mining applications implement systems with heterogeneous classifiers and adapt online to its input data stream characteristics  ...  In [12] , a VLSI implementation of the real-time learning and recognition system based on adaptive K-Means learning algorithm has been implemented on FPGA.  ... 
doi:10.1109/isvlsi.2014.77 dblp:conf/isvlsi/KanounRAS14 fatcat:vu3tpx4wffa6fc5kiteuf6gnxa

Error-adaptive classifier boosting (EACB): Exploiting data-driven training for highly fault-tolerant hardware

Zhuo Wang, Robert Schapire, Naveen Verma
2014 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)  
The approach exploits machine learning for modeling not only complex sensor signals but also error manifestations due to hardware faults.  ...  Adaptive boosting is exploited in the architecture for performing iterative data-driven training.  ...  Then AdaBoost is introduced, which we will exploit in a machine-learning kernel that is itself allowed to be highly fault prone, substantially expanding the concept of DDHR.  ... 
doi:10.1109/icassp.2014.6854329 dblp:conf/icassp/WangSV14 fatcat:be52avjoefcn3f4jokdpoibfza

Exploiting Parallelism Inherent in AIRS, an Artificial Immune Classifier [chapter]

Andrew Watkins, Jon Timmis
2004 Lecture Notes in Computer Science  
The work in this paper presents the first steps at realizing a parallel artificial immune system for classification.  ...  The mammalian immune system is a highly complex, inherently parallel, distributed system.  ...  Conclusion Our goal with this study was to explore ways of exploiting parallelism inherent in an artificial immune system for decreased overall runtime.  ... 
doi:10.1007/978-3-540-30220-9_34 fatcat:crcprhcedbelzfmfr5mfqiwbl4
« Previous Showing results 1 — 15 out of 116,492 results