Malware Classification with Deep Convolutional Neural Networks

Kalash, Mahmoud; Rochan, Mrigank; Mohammed, Noman; Bruce, Neil D. B.; Wang, Yang; Iqbal, Farkhund

doi:10.1109/ntms.2018.8328749

Cited by 273 publications

(157 citation statements)

References 13 publications

Supporting

Mentioning

155

Contrasting

Unclassified

Order By: Relevance

“…These developments have allowed deep learning to progress from research to industrial applications. Recently, it has been shown to have comparable performance to human experts [52][53][54]. For PM 2.5 prediction with deep learning, there have been some attempts with different structures, including a deep neural network, long and short term memory (LSTM), and a convolutional neural network (CNN) [33,35,37,38,55,56].…”

Section: Deep Learningmentioning

confidence: 99%

PM2.5 Prediction Based on Random Forest, XGBoost, and Deep Learning Using Multisource Remote Sensing Data

et al. 2019

View full text Add to dashboard Cite

In recent years, air pollution has become an important public health concern. The high concentration of fine particulate matter with diameter less than 2.5 µm (PM2.5) is known to be associated with lung cancer, cardiovascular disease, respiratory disease, and metabolic disease. Predicting PM2.5 concentrations can help governments warn people at high risk, thus mitigating the complications. Although attempts have been made to predict PM2.5 concentrations, the factors influencing PM2.5 prediction have not been investigated. In this work, we study feature importance for PM2.5 prediction in Tehran’s urban area, implementing random forest, extreme gradient boosting, and deep learning machine learning (ML) approaches. We use 23 features, including satellite and meteorological data, ground-measured PM2.5, and geographical data, in the modeling. The best model performance obtained was R2 = 0.81 (R = 0.9), MAE = 9.93 µg/m3, and RMSE = 13.58 µg/m3 using the XGBoost approach, incorporating elimination of unimportant features. However, all three ML methods performed similarly and R2 varied from 0.63 to 0.67, when Aerosol Optical Depth (AOD) at 3 km resolution was included, and 0.77 to 0.81, when AOD at 3 km resolution was excluded. Contrary to the PM2.5 lag data, satellite-derived AODs did not improve model performance.

show abstract

Section: Deep Learningmentioning

confidence: 99%

PM2.5 Prediction Based on Random Forest, XGBoost, and Deep Learning Using Multisource Remote Sensing Data

et al. 2019

View full text Add to dashboard Cite

show abstract

“…Recently, an increasing number of contemporary methods using malware image to classify turn to employ deep learning models like CNN due to their powerful extraction ability, resistance to noise and robustness when processing different modalities of data [3,28,30,31]. VGG-based and ResNet-based advanced architectures are also exploited [26,[32][33][34]. Differently, Vu et al [35] developed a novel approach using hybrid transformation to convert malware to color images that convey malware semantics to perform malware classification tasks.…”

Section: Malware Imagementioning

confidence: 99%

ConvProtoNet: Deep Prototype Induction towards Better Class Representation for Few-Shot Malware Classification

Tang

Wang

2020

Applied Sciences

View full text Add to dashboard Cite

Traditional malware classification relies on known malware types and significantly large datasets labeled manually which limits its ability to recognize new malware classes. For unknown malware types or new variants of existing malware containing only a few samples each class, common classification methods often fail to work well due to severe overfitting. In this paper, we propose a new neural network structure called ConvProtoNet which employs few-shot learning to address the problem of scarce malware samples while prevent from overfitting. We design a convolutional induction module to replace the insufficient prototype reduction in most few-shot models and generates more appropriate class-level malware prototypes for classification. We also adopt meta-learning scheme to make classifier robust enough to adapt unseen malware classes without fine-tuning. Even in extreme conditions where only 5 samples in each class are provided, ConvProtoNet still achieves more than 70% accuracy on average and outperforms other traditional malware classification methods or existed few-shot models in experiments conducted on several datasets. Extra experiments across datasets illustrate that ConvProtoNet learns general knowledge of malware which is dataset-invariant and careful model analysis proves effectiveness of ConvProtoNet in few-shot malware classification.

show abstract

“…2, The input layer has only one node as the frequency, 16 nodes inside the hidden layer to add a non-linearity to model and 6 nodes in the output layer as the representation of the ground plane height, triangle side length, circle diameters, return loss, gain and directivity of the antenna [14] [15]. The algorithm was experimented using a different-numbers of hidden layers and neurons [16] [17]. However, the architecture explained above is the best possible combination and has the best result in minimizing the loss and converge faster than any other combination.…”

Section: Network Achitecturementioning

confidence: 99%

Performance Tuning of Spade Card Antenna using Mean Average Loss of Backpropagation Neural Network

Mujahidin¹,

Prasetya²,

Nachrowie³

et al. 2020

IJACSA

View full text Add to dashboard Cite

The microstrip antennas have different dimensions to get the desired performance, especially for microstrip antennas that have complex components and dimensions with the performance: the range of frequency at 2.4 GHz until 3.6 GHz, Maximum Power of Gain value is 5.83 dB and the minimum value is 3 dB and Maximum Directivity Value is 6.22 and the minimum value is 3.32. in consequence, needs to fill the demand for a new and the corresponding design as solvent to adaptive matching as tuner the frequency on antenna design that needs requires a complex mathematical method and simulation. This paper has the novel design to tune the performance of spade card microstrip antenna that can operate on the single, dual or multiband and able to produce circular or linear polarization using Backpropagation Neural Network in order to obtain an optimum design with a backpropagation algorithm as a solution to simplify the design process. As a result, after 20000 epochs the training loss is around 0.044 and the testing loss is around 0.058. The model has a good performance despite only using a few numbers of training data.

show abstract

Malware Classification with Deep Convolutional Neural Networks

Cited by 273 publications

References 13 publications

PM2.5 Prediction Based on Random Forest, XGBoost, and Deep Learning Using Multisource Remote Sensing Data

PM2.5 Prediction Based on Random Forest, XGBoost, and Deep Learning Using Multisource Remote Sensing Data

ConvProtoNet: Deep Prototype Induction towards Better Class Representation for Few-Shot Malware Classification

Performance Tuning of Spade Card Antenna using Mean Average Loss of Backpropagation Neural Network

Contact Info

Product

Resources

About