Federated Learning for Malware Detection
Abstract (300 words maximum)
In this era of computers, we encounter many problems when handling the devices around us. One of the major problems faced is malware detection. Malware is malicious software that enters a device without the user's permission and accesses sensitive and confidential data. Malware is diverse, and This makes malware detection a challenging task as it does not have a global database of malware to identify. This paper discusses traditional non-machine learning-based and machine learning-based malware detection techniques compared to the federated learning-based architecture. The non-machine learning-based malware detection is based on permissions and signatures. In comparison, machine learning-based malware detection happens using machine learning-based algorithms. This comparison talks about an efficient way to detect malware. The Federated Learning-based Malware detection process is an advanced concept of machine learning, which happens through a decentralized training model derived from different devices to form a global model. In this process, individually trained models from various devices will be collected and analyzed to develop a distributed model for malware detection. Furthermore, this paper explains the advantages and disadvantages of using Federated Learning to detect malware in different areas of use.
Academic department under which the project should be listed
CCSE - Information Technology
Primary Investigator (PI) Name
Liang Zhao
Federated Learning for Malware Detection
In this era of computers, we encounter many problems when handling the devices around us. One of the major problems faced is malware detection. Malware is malicious software that enters a device without the user's permission and accesses sensitive and confidential data. Malware is diverse, and This makes malware detection a challenging task as it does not have a global database of malware to identify. This paper discusses traditional non-machine learning-based and machine learning-based malware detection techniques compared to the federated learning-based architecture. The non-machine learning-based malware detection is based on permissions and signatures. In comparison, machine learning-based malware detection happens using machine learning-based algorithms. This comparison talks about an efficient way to detect malware. The Federated Learning-based Malware detection process is an advanced concept of machine learning, which happens through a decentralized training model derived from different devices to form a global model. In this process, individually trained models from various devices will be collected and analyzed to develop a distributed model for malware detection. Furthermore, this paper explains the advantages and disadvantages of using Federated Learning to detect malware in different areas of use.