Claim Missing Document
Check
Articles

Found 2 Documents
Search
Journal : Building of Informatics, Technology and Science

Eksperimen Pengujian Optimizer dan Fungsi Aktivasi Pada Code Clone Detection dengan Pemanfaatan Deep Neural Network (DNN) Errissya Rasywir; Yovi Pratama; Fachruddin Fachruddin
Building of Informatics, Technology and Science (BITS) Vol 4 No 2 (2022): September 2022
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v4i2.1776

Abstract

The problem of similarity (similarity) of program code can be one solution to the plagiarism detection approach. Plagiarism raises a form of action and consequences of plagiarism itself if the source used is not open source. Plagiarism is an act of deception of the work of others without the knowledge of the original author, which violates a Copyright and Moral Rights. With the increasing amount of data and data complexity, deep learning provides solutions for predictive analytics, with increased processing capabilities and optimal processor utilization. Deep learning shows success and improves the classification model in this field. On the other hand, clone detection code with massive, varied and high-speed data volumes requires feature extraction. With the potential of deep learning to extract better features, deep learning techniques are suitable for code clone detection. For this reason, it is necessary to develop a clone detection code that can process data from a programming language by utilizing deep learning. Based on the results of experiments conducted on 100 PHP program code data files, experimented with several types of activation function and optimizer methods. The average value of the resulting accuracy is good. With a variety of activation functions that we use such as Relu, Linear, Sigmoid, Softmax, Tanh, Elu, Selu, Softplus, Softsign, hard, and sigmoid, as well as variations of the optimizer used are Adagrad, RMSProp, SGD, Adadelta, Adam, Adamax and Nadam , the best attribute selection is in the Selu function and the RMSProp optimizer. The number of epochs used is 1000, the number of neurons per layer is 500 and the best number of hidden layers is 10, the average accuracy is 0.900
Reduksi False Positive Pada Klasifikasi Job Placement dengan Hybrid Random Forest dan Auto Encoder Pahlevi, M. Riza; Rasywir, Errissya; Pratama, Yovi; Istoningtyas, Marrylinteri; Fachruddin, Fachruddin; Yaasin, Muhammad
Building of Informatics, Technology and Science (BITS) Vol 5 No 4 (2024): March 2024
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v5i4.4864

Abstract

The False Positive (FP) interpretation shows a negative prediction result and is a type 1 error answer with an incorrect positive prediction result. Based on this, we try to reduce type 1 errors to increase the accuracy value of the classification results. A low FP rate is critical for the use of Computer Aided Detection (CAD) systems. In this research proposal, to reduce FP, we use a Random Forest (RF) evaluation result design which will be reinterpreted by the Auto Encoder (AE) algorithm. The RF algorithm was chosen because it is a type of ensemble learning that can optimize accuracy in parallel. RF was chosen because it performs bagging on all Decision Tree (DT) outputs used. To suppress TP reduction more strongly, we use the Auto Encoder (AE) algorithm to reprocess the class bagging results from RF into input in the AE layer. AE uses reconstruction errors, which in this case is Job Placement classification. From the test results, it was found that combining the use of a random forest using C4.5 as a decision tree with an Autoencoder can increase accuracy in the Job Placement Classification task by a difference of 0.004652 better than without combining it with an autoencoder. Apart from that, in testing using a combination of RF and AE, fewer False Positive (FP) values ​​were produced, namely 11 items in the Cross Validation-5 (CV-5) Test, then 13 items in the Cross Validation-10 (CV-10) test and in testing split training data of 60%, the FP was only 12. This value is less than the false positives produced by testing without Autoencoder, namely 12 items on CV-5, 15 items on CV-10, and 13 on split training data