Evaluation of Deep Learning Models for Continuous Authentication Using Behavioral Biometrics

This article describes a multimodal framework based on behavioral biometrics and location for continuous user authentication in mobile banking applications. The proposed framework joins the typing and screen slide behavior patterns with the GPS location of the user in the authentication process. During the performance evaluation experiments, we used more than 6 different machine learning algorithms. Main results show an average accuracy between 78 and 91%.

In cybersecurity, the rise of fileless malware poses a significant challenge to endpoint security. Traditional detection methods often fail against these sophisticated attacks, necessitating advanced techniques like deep learning models. This study highlights the limitations of Bi-Directional Long Short-Term Memory (BLSTM) models in dynamic malware analysis and proposes enhancements through Convolutional Long Short-Term Memory (ConvLSTM) architecture. BLSTM models process input sequences in forward and backward directions, combining the results into one output. While this dual-layer approach improves analysis, it is time-consuming, potentially increasing the risk of fileless malware attacks. A key limitation of BLSTM is the lack of parameter sharing between forward and backward directions. This reduces its ability to capture spatial and temporal features simultaneously, hindering effectiveness in detecting fileless malware. To address this, the ConvLSTM model consolidates feature extraction within a single LSTM cell layer. ConvLSTM breaks down samples into subsequence and uses timesteps for additional feature extraction, enabling spatial-temporal data analysis and improving malware prediction accuracy. The model was tested using a dynamic malware dataset. Unlike traditional LSTM, ConvLSTM integrates convolutional layers, allowing parameter sharing across both spatial and temporal dimensions. This reduces computational complexity and improves model performance in handling multidimensional data. The research re-simulated prior work with BLSTM using the same malware dataset. The Spyder app ran the event simulator and the ConvLSTM model's results replaced BLSTM's using identical parameters. Time, accuracy and loss were the main performance metrics. ConvLSTM outperformed BLSTM, achieving 98% detection accuracy compared to BLSTM's 90%. It also significantly reduced processing time, averaging 10 seconds, while BLSTM took 22 seconds. ConvLSTM experienced lower losses, averaging 10% per epoch versus BLSTM's 20%. In conclusion, ConvLSTM offers superior performance over BLSTM in fileless malware detection. Its enhanced computational efficiency and ability to quickly mitigate threats make it a robust solution for fortifying endpoint security against evolving cyber threats. ConvLSTM holds potential in strengthening defence mechanisms against sophisticated malware attacks, providing a proactive approach to safeguarding networks and data.

This paper presents an overview on the possible use of behavioral biometrics methods in mobile banking and payment applications. As mobile applications became more common, more and more users conduct payments using their smartphones. While requiring secure services, the customers often do not lock their devices and expose them to potential misuse and theft. Banks and financial institutions apply multiple anti-fraud and authentication systems - but to ensure the required usability, they must develop new ways to authenticate their users and authorize transactions. Answer to this problem comes with a family of behavioral biometric methods which can be utilized to secure those applications without hindering the usability. The goal of this paper is to describe potential areas in which behavioral biometrics can be used to ensure more secure mobile payments, increase usability and prevent frauds.

This research is aimed to examine the gender difference toward intention to use the mobile banking application in Semarang. Data is collected by self - administered questionnaire, and pass over to bank customer which done their financial transaction by Automatic Teller Machine. The Sample used in this research is 50 respondents, consist of 30 male respondents and 20 female respondents who never have a mobile banking application before. The results of this study indicate that there is a gender difference in the intention to use mobile banking applications. On the other side, even it is easy to be used, female customers are not feeling the benefit offered, thus resistant to use mobile banking applications. This result brings another contradiction toward The Technological Acceptance Model. Data is examined by the SPSS application, and Mann – Whitney statistical model.Keywords: Mobile application; bank; gender difference

This research is a study to analyze the effect of Interface Design on customer loyalty in mobile banking applications in Indonesia. This research methodology uses a questionnaire distribution technique for millennial generation respondents who use mobile banking in Indonesia, using the Likert Scale method, and data processing techniques using the Structural Equation Modeling (SEM) method. This research aims to examine the significant influence of Interface Design on customer loyalty in mobile banking applications in Indonesia. This digital transformation in the banking industry provides benefits for both banks and their customers. Banking digitalization can provide benefits for accelerating customer transactions, avoiding human errors for manual transactions at offline branches, and increasing banks’ efficiency because it reduces manual processes in serving customer transactions. For customers, mobile banking applications accelerate transactions, time efficiency without having to come to an offline branch, and flexibility to create transactions. The hypothesis indicates that interface design significantly impacts loyalty customers using mobile banking applications. This research is useful for the banking industry to accommodate a user-friendly interface design to increase the competitive advantage for banks that provide mobile banking applications.

Mobile banking, sometimes referred to as M-Banking, Mbanking or SMS Banking, is a term used for performing balance checks, account transactions, payments, credit applications and other banking transactions through a mobile device such as a mobile phone or Personal Digital Assistant (PDA). Mobile banking has until recently most often been performed via SMS or the Mobile Web. Apple's initial success with iPhone and the rapid growth of phones based on Google's Android (operating system) have led to increasing use of special client programs, called apps, downloaded to the mobile device hence increasing the number of banking applications that can be made available on mobile phones . This in turn has increased the popularity of mobile device use in regards to personal banking activities. Due to the characteristics of wireless medium, limited protection of the nodes, nature of connectivity and lack of centralized managing point, wireless networks tend to be highly vulnerable and more often than not they become subjects of attack. This paper proposes to identify potential threats associated with communication between a mobile device and the back end server in mobile banking applications. The paper should be able to identify the techniques associated with Man in the middle attacks during communication between a mobile device and a back end server and propose controls that will ensure that data theft does not occur during such sessions.

Continuous authentication with feature-level fusion of touch gestures and keystroke dynamics to solve security and usability issues

PurposeThe research aims to addresses the limitations of previous literature regarding choosing the appropriate conceptualization of trust (i.e. interpersonal trust or system trust) and the role of design aesthetics in generating system trust and intention to adopt mobile banking.Design/methodology/approachThe research conducts two studies. Study 1 determines the degree of humanness in a mobile banking application. Study 2 tests the research model. A total of 261 respondents participate in Study 1 and 491 in Study 2. Statistical Package for the Social Sciences (SPSS) (study 1) and SmartPLS (PLS software) (study 2) are used to test the hypotheses.FindingsStudy 1 establishes that the mobile banking application is perceived to have low humanness. Thus, it is expected that system trust is more appropriate to use than interpersonal trust. Study 2 demonstrates that (i) system trust is more useful than interpersonal trust in the mobile banking context and (ii) design aesthetics positively influences consumer system trust and intention to adopt.Originality/valueTo the best of the authors' knowledge, the current research is the first to distinguish empirically between system trust and interpersonal trust and identify the best choice of mobile banking trust type. Specifically, the study determined the choice of system trust for mobile banking through a priori humanness measures and validated this choice by measuring both system trust and interpersonal trust, which has not been done before. In addition, retail banking should consider the influence of design aesthetics on consumer trust and incorporate elements that enhance consumers' opinions about the mobile banking application's functionality, reliability and helpfulness.

This study aims to find out how mobile banking is a communication medium for banking services at Bank BTN Bengkulu Branch, using Lasswell theory in Kurniawan (2018:62) which consists of: Who (who is the communicator), say what (what message he states), in which channel (what channel/media is used), to whom (who is the communicant) and whit what effect (the expected effect). This research is a qualitative research with a descriptive design. Data collection techniques in this study are primary data and secondary data. The results show that Who (who is the communicator), the communicator in this study is a mobile banking application at Bank BTN Bengkulu Branch. Say What (what message it states), the message given by the mobile banking application in this study is in the form of sentences or words and numbers desired by customers as mobile banking users of Bank BTN Bengkulu Branch. In Which Channel (What channel/media is used), the media in this study is the media used by customers in the use of mobile banking at Bank BTN Bengkulu Branch, namely the andorid phone. To Whom (Who is the communicant), the communicant in this study is the customer of Bank BTN Bengkulu Branch. Whit What Effect (Expected Effect), the effect in this study is to direct the communicator (application) to communicate according to what the customer wants so that ongoing transactions through the mobile banking application can run successfully and successfully.

Session authentication schemes establish the identity of the user only at the beginning of the session, so they are vulnerable to attacks that tamper with communications after the establishment of the authenticated session. Moreover, smartphones themselves are used as authentication means, especially in two-factor authentication schemes, which are often required by several services. Whether the smartphone is in the hands of the legitimate user constitutes a great concern, and correspondingly whether the legitimate user is the one who uses the services. In response to these concerns, Behavioral Biometrics (BB) Continuous Authentication (CA) technologies have been proposed on a large corpus of literature. This paper presents a research on the development and validation of a BBCA system (named BioPrivacy), that is based on the user’s keystroke dynamics, using a Multi-Layer Perceptron (MLP). Also, we introduce a new behavioral biometrics collection tool, and we propose a methodology for the selection of an appropriate set of behavioral biometrics. Our system achieved 97.18% Accuracy, 0.02% Equal Error Rate (EER), 97.2% True Acceptance Rate (TAR) and 0.02% False Acceptance Rate (FAR).KeywordsMachine learningBehavioral biometricsContinuous authenticationMobile devicesMulti-layer perceptron (MLP)

Behavioral biometrics & continuous user authentication on mobile devices: A survey

Security is a crucial aspect in mobile banking applications, given the increasing cyber threats that can harm both users and financial institutions. The implementation of biometric-based security systems, such as fingerprint scanning and facial recognition, offers a more secure solution than traditional authentication methods. This article discusses the implementation of biometric security systems in mobile banking applications, evaluating its effectiveness and challenges in its integration. Using literature review and system prototyping methods, this research shows that biometrics can increase the level of security and convenience for users, but requires attention to privacy issues and technology compatibility.

The rapid growth of mobile banking applications has transformed the financial landscape, providing convenience and accessibility to users worldwide. All 11 commercial banks in Albania provide mobile banking options to their customers. Smartphones have become the primary platform for digital interactions, with mobile banking emerging as the preferred method for over 73% of online banking users in the country. Recent literature suggests that user-friendly features, ease of navigation, simple language, continuous updates, robust safety measures, and active reviewer engagement contribute significantly to app downloads, ongoing usage, conversion rates, and customer satisfaction. Ultimately, these factors enhance the efficiency of banking operations. This study analyses reviews of Albanian mobile banking applications on the Google Play Store to identify trends and patterns. The findings reveal a correlation between bank size, mobile banking app ratings, and reviewer engagement. Extending the analysis to Western Balkan banks confirms a similar bank size, app rating, and user engagement tendency. This work underscores the necessity of a quality assurance system to effectively apply customer feedback for useful mobile banking application updates.

Mobile devices allow users to access applications anytime and anywhere. Through mobile devices, users can download and get access to banking services, using mobile banking applications. In order to perform banking activities, users need to trust in this service first, and this includes the quality of the Internet connection. Moreover, considering that mobile devices have different screen sizes, these differences may also affect the intention to use mobile banking. Therefore, the aim of this paper was to analyze the potential effect of the size of mobile devices and of the mobile Internet data services on trust in mobile banking. The quantitative analysis includes descriptive statistics, confirmatory factor analysis and structural equation modeling. The main results suggest that mobile Internet data services presented a positive effect on trust in mobile banking, but the type of mobile device has no significant effect on trust. The discussion of the results considers the possibility of strategic alliances between different sectors to improve the adoption of mobile banking in developing countries.

Nowadays, most smartphones feature a number of strong sensors, such as direction, network, location, and motion sensors. Motion and inertial sensors (i.e., accelerometer, gyroscope, etc.) are particularly popular in human activity recognition (HAR) for detecting human physical activity which helps in many research areas of the Internet of healthcare things (IoHT) including several patient’s diseases such as Parkinson, obesity, cardiovascular, and diabetes. Deep learning (DL) techniques have been increasingly popular over the years as a result of their significant outcomes. The use of deep learning techniques to recognize human physical activity in wearable and mobile sensor situations has also gotten a lot of attention from around the world. In this paper, a deep neural network (DNN) combining bidirectional long short-term memory (Bi-LSTM) and convolutional neural network (CNN) is proposed. The performance of the model has been evaluated on two publicly available datasets: WISDM and UCI-HAR. The model has achieved 97.96 and 97.15% accuracy for the WISDM and UCI-HAR, respectively. Moreover, the simulation results show the effectiveness of the proposed work compared to other state-of-the-art methods.KeywordsHuman activity recognitionDeep learningIoHTConvolutional neural networks (CNN)Bidirectional long short-term memory networks (Bi-LSTM)