Filters








14,512 Hits in 3.4 sec

Learning Execution through Neural Code Fusion [article]

Zhan Shi, Kevin Swersky, Daniel Tarlow, Parthasarathy Ranganathan, Milad Hashemi
2020 arXiv   pre-print
While there is a growing body of work on using Graph Neural Networks (GNNs) to learn representations of source code, these representations do not understand how code dynamically executes.  ...  In this work, we propose a new approach to use GNNs to learn fused representations of general source code and its execution.  ...  We call our approach neural code fusion (NCF).  ... 
arXiv:1906.07181v2 fatcat:czxoxlbifbgn3kvssezsyzjclu

Accelerating Deep Learning Inference with Cross-Layer Data Reuse on GPUs [article]

Xueying Wang, Guangli Li, Xiao Dong, Jiansong Li, Lei Liu, Xiaobing Feng
2020 arXiv   pre-print
To achieve the balance between computation and memory access, we explore the fusion opportunities in the CNN computation graph and propose three fusion modes of convolutional neural networks: straight,  ...  Accelerating the deep learning inference is very important for real-time applications.  ...  TVM [2] is a compiler to generate portable code for deep learning applications across diverse hardware platforms.  ... 
arXiv:2007.06000v2 fatcat:nwv6glfp4ndabm5xuuhmnv5zqu

Planning with iFALCON: Towards A Neural-Network-Based BDI Agent Architecture

Budhitama Subagdja, Ah-Hwee Tan
2008 2008 IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology  
This paper shows that a simplified plan representation can be encoded as weighted connections in the neural network through a process of supervised learning.  ...  Based on multichannel network model called fusion ART, iFALCON is developed to bridge the gap between a self-organizing neural network that autonomously adapts its knowledge and the BDI agent model that  ...  . fusion ART is a neural architecture that unifies a number of neural network designs, most notably ART [3, 2] , Adaptive Resonance Associative Map (ARAM) [14] and Fusion Architecture for Learning,  ... 
doi:10.1109/wiiat.2008.29 dblp:conf/iat/SubagdjaT08 fatcat:hkabflyo25c45kfzesidkzzcoi

A Study on the Prediction of Program Complexity Section for Offloading Execution Decision

Jaehyun Kim, Yangsun Lee
2019 International Journal of Control and Automation  
The program complexity area prediction predicts the execution complexity of the program by learning the program complexity estimate and actual execution complexity analyzed by the static profiler.  ...  In this paper, we predicted a program area complexity based on deep learning to solve this problem.  ...  Since the program has execution order, this paper adopts LSTM neural network as a deep learning neural network.  ... 
doi:10.33832/ijca.2019.12.8.10 fatcat:juw32vmvbfg2niwgb4wjvcfnky

Hybrid Deep Neural Network based Performance Estimation Method for Efficient Offloading on IoT-Cloud Environments

Yunsik Son, Seman Oh, Yangsun Lee
2018 International Journal of Grid and Distributed Computing  
In this paper, CPU utilization trend, which is one of the workload indices, is predicted through deep learning in order to decide offloading execution considering the workload of the IoT devices and clouds  ...  In this paper, we present four CPU usage models and introduce a technique for predicting server load based on hybrid deep neural network.  ...  Acknowledgment This research was supported by Basic Science Research Program through the National Research Foundation of Korea(NRF) funded by the Ministry of Science, ICT and Future Planning (No. 2016R1A2B4008392  ... 
doi:10.14257/ijgdc.2018.11.7.03 fatcat:cfle64i5vvecpfgscvcabkhqey

A Biologically-Inspired Cognitive Agent Model Integrating Declarative Knowledge and Reinforcement Learning

Ah-Hwee Tan, Gee-Wah Ng
2010 2010 IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology  
By replacing the production system of ACT-R by a fusion ART model, FALCON-X integrates high-level deliberative cognitive behaviors and real-time learning abilities, based on biologically plausible neural  ...  networks called fusion Adaptive Resonance Theory (fusion ART).  ...  [3] and the fusion Adaptive Resonance Theory (fusion ART) neural model [5] .  ... 
doi:10.1109/wi-iat.2010.210 dblp:conf/iat/TanN10 fatcat:6jy5zcfucfdsbozlwfxnoc56pu

Learning-Based Detection for Malicious Android Application Using Code Vectorization

Lin Liu, Wang Ren, Feng Xie, Shengwei Yi, Junkai Yi, Peng Jia, Leandros Maglaras
2021 Security and Communication Networks  
This method extracts the static features in the core code of the Android application by decompiling APK files, then performs code vectorization processing, and uses the deep learning network for classification  ...  In this paper, a malicious Android application detection method is proposed, which is implemented by the deep network fusion model.  ...  for deep learning than the dynamic files generated by the application running in the sandbox. e Android application is developed in the Java language, and its code is executed under the interpretation  ... 
doi:10.1155/2021/9964224 fatcat:c42il3zasbagrlkazpi4xniuvy

Latte: a language, compiler, and runtime for elegant and efficient deep neural networks

Leonard Truong, Rajkishore Barik, Ehsan Totoni, Hai Liu, Chick Markley, Armando Fox, Tatiana Shpeisman
2016 Proceedings of the 37th ACM SIGPLAN Conference on Programming Language Design and Implementation - PLDI 2016  
Using networks described using Latte, we demonstrate 3-6× speedup over Caffe (C++/MKL) on the three state-of-the-art ImageNet models executing on an Intel Xeon E5-2699 v3 x86 CPU.  ...  Deep neural networks (DNNs) have undergone a surge in popularity with consistent advances in the state of the art for tasks including image recognition, natural language processing, and speech recognition  ...  ) Figure 12 : 12 The final code after tiling, fusion, and parallelization.  ... 
doi:10.1145/2908080.2908105 dblp:conf/pldi/TruongBTLMFS16 fatcat:phob6d5p4nb55cx4l6vok4y6y4

Agent-Augmented Co-Space: Toward Merging of Real World and Cyberspace [chapter]

Ah-Hwee Tan, Yilin Kang
2010 Lecture Notes in Computer Science  
Following the notion of embodied intelligence, we propose to develop cognitive agents, based on a family of self-organizing neural models, known as fusion Adaptive Resonance Theory (fusion ART).  ...  Through realistic 3D modelling and animation technologies, Co-Space simulates the real world in terms of look-and-feel of our physical surrounding.  ...  The fusion ART pattern processing cycle comprises five key stages, namely code activation, code competition, activity readout, template matching, and template learning, as described below.  ... 
doi:10.1007/978-3-642-16576-4_22 fatcat:tw3b6hjzpjfedmmet527vjpfoi

The rise of machine learning for detection and classification of malware: Research developments, trends and challenges

Daniel Gibert, Carles Mateu, Jordi Planes
2020 Journal of Network and Computer Applications  
This survey aims at providing a systematic and detailed overview of machine learning techniques for malware detection and in particular, deep learning techniques.  ...  the challenges and limitations of traditional machine learning and (3) it analyzes recent trends and developments in the field with special emphasis on deep learning approaches.  ...  Packers hide the real code of a program through one or more layers of compression. Then, at runtime the unpacking routines restore the original code in memory and execute it.  ... 
doi:10.1016/j.jnca.2019.102526 fatcat:3bf6afjqpnb53eoeghfxjeaus4

SparseDNN: Fast Sparse Deep Learning Inference on CPUs [article]

Ziheng Wang
2021 arXiv   pre-print
The last few years have seen gigantic leaps in algorithms and systems to support efficient deep learning inference.  ...  To tackle this challenge, we present SparseDNN, a sparse deep learning inference engine targeting CPUs.  ...  We show that through both network-level and kernel-level optimizations, we can significantly speed up the execution of unstructured sparse neural networks on modern datacenter CPUs where most deep learning  ... 
arXiv:2101.07948v4 fatcat:bs6rdifdlvat3hr4n435w65h2y

Malicious Code Invariance Based On Deep Learning

2021 International Journal of Information Technology Infrastructure  
The malicious code detection can be possibly by using convolutional neural network (CNN).Themalicious code can be categorized in to different families.  ...  The neural network architecture classifies the malicious code based on the collected dataset. The dataset contains different families of malicious code.  ...  CONCLUSION The malicious code invariance based on deep learning proposed a novel method to improve the detection of malware variants through the application of deep learning along with convolutional neural  ... 
doi:10.30534/ijiti/2021/011032021 fatcat:gslsmvwa7vezjhqqqetsttpfqa

Real-Time Execution of Large-scale Language Models on Mobile [article]

Wei Niu, Zhenglun Kong, Geng Yuan, Weiwen Jiang, Jiexiong Guan, Caiwen Ding, Pu Zhao, Sijia Liu, Bin Ren, Yanzhi Wang
2020 arXiv   pre-print
We propose the first compiler-aware neural architecture optimization framework.  ...  Our framework can guarantee the identified model to meet both resource and real-time specifications of mobile devices, thus achieving real-time execution of large transformer-based models like BERT variants  ...  Neural Architecture Search With the development of AI democratization, automatic machine learning (AutoML) has been a hot research area in the past few years.  ... 
arXiv:2009.06823v2 fatcat:47xsbdnyifgv3loqqdtaarc46u

Integration of Static and Dynamic Analysis for Malware Family Classification with Composite Neural Network [article]

Yao Saint Yen, Zhe Wei Chen, Ying Ren Guo, Meng Chang Chen
2019 arXiv   pre-print
Deep learning has been used in the research of malware analysis.  ...  In this paper, we combine static and dynamic analysis features with deep neural networks for Windows malware classification.  ...  In early fusion neural network 1, the features are concatenated as a vector paaed through a fully connected layer for malware classification.  ... 
arXiv:1912.11249v1 fatcat:sz5fndjgiffn3b5y2aiafjdcua

A Vulnerability Detection System Based on Fusion of Assembly Code and Source Code

Xingzheng Li, Bingwen Feng, Guofeng Li, Tong Li, Mingjin He, Liguo Zhang
2021 Security and Communication Networks  
Second, these slices are aligned by the proposed code alignment algorithm. Third, aligned code slices are converted into vector and input into a hyper fusion-based deep learning model.  ...  This paper implements a vulnerability detection system by combining source code and assembly code models. First, code slices are extracted from the source code and assembly code.  ...  Conclusion is paper proposes a system for detecting vulnerabilities in software through deep learning.  ... 
doi:10.1155/2021/9997641 fatcat:4h2w33ehuzendnbljhbrgh5sju
« Previous Showing results 1 — 15 out of 14,512 results