| Autor: |
Ram Hari Regmi, Arun K. Timalsina |
| Rok vydání: |
2018 |
| Předmět: |
|
| Zdroj: |
2018 IEEE 3rd International Conference on Computing, Communication and Security (ICCCS). |
| DOI: |
10.1109/cccs.2018.8586834 |
| Popis: |
Increasing trade volume adds up various challenges and risks for customs to maintain balance between trade facilitation and strong border control. With limited resources and manpower, it’s quite difficult to have exhaustive physical examination of all import and export consignments. To balance control and facilitation Revised Kyoto Convention (RKC) and World Trade Organization (WTO) Trade Facilitation Agreement (TFA) have clearly stated about implementation of effective risk management system. In this paper, deep learning model was trained and tested to segregate high risk and low risk consignment on randomly selected 200,000 data from Nepal Customs of the year 2017. Model was tested using supervised learning utilizing inspection result provided by Nepal Customs. Deep learning has improved accuracy and seizure rate than that of decision Tree (DT) and Support Vector Machine (SVM). All three methods have achieved a better result than current rule based risk management system. ANN had achieved better result than DT and SVM, by achieving 81% of seizure rate under 9% inspection. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
