Tai-Hung Lai
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Tai-Hung Lai has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Gang-Cheng Huang
China University of Technology
3 shared publications - Ko-Chin Chang
Department of Electrical and Electronic Engineering, Chung Cheng Institute of Technology, National Defense University, Taoyuan 335009, Taiwan
2 shared publications - Chao-Lung Chou
Department of Information Engineering and Computer Science, Feng Chia University, Taichung 40724, Taiwan
1 shared publication - Chiang-Lung Liu
National Defense University
1 shared publication Ko-Chin Chang
Department of Electrical and Electronic Engineering, Chung Cheng Institute of Technology, National Defense University, Taoyuan County, Taiwan, ROC
1 shared publicationWen-Bin Lin
School of Defense Science, Chung Cheng Institute of Technology, National Defense University, Taoyuan County, Taiwan, ROC
1 shared publication
Latest Publications
- Open Access
- 23 November 2025
- Published by MDPI AG in Mathematics
- Vol. 13 (23) , 3753
Improving the Performance of Static Malware Classification Using Deep Learning Models and Feature Reduction Strategies
The rapid evolution of malware continues to pose severe challenges to cybersecurity, highlighting the need for accurate and efficient detection systems. Traditional signature- and heuristic-based methods are increasingly inadequate against sophisticated threats, which has motivated the use of machine learning and deep learning for static malware classification. In this study, we propose three deep neural network (DNN) architectures tailored for the binary classification of Portable Executable (PE) files. The models were trained and validated on the EMBER 2017 dataset and further tested on the independent REWEMA dataset to evaluate their cross-dataset generalization capabilities. To address the computational burden of high-dimensional feature vectors, two feature reduction strategies were examined: the Kumar method, which selected 276 features, and the LightGBM-based intersection method, which identified 206 shared features. Experimental results showed that the proposed Model III consistently achieved the best overall performance, outperforming LightGBM (v3.3.5) and the other DNN models in terms of accuracy, recall, and F1-score. Notably, its recall exceeded that of LightGBM by 0.73%, highlighting its superiority in reducing false negative rates. Feature reduction further demonstrated that significant dimensionality reduction could be achieved without compromising classification quality, with the Kumar method achieving the best balance between accuracy and efficiency. Cross-dataset validation revealed performance degradation across all models due to distributional shifts, but the decline was less significant for the DNNs, confirming its greater adaptability compared with LightGBM. These findings demonstrate that architectural optimization and appropriate feature selection can significantly improve the performance of static malware classification. This study also provides empirical benchmarks and methodological guidance for developing accurate, efficient, and resilient malware detection systems that are resilient to evolving threats.
- 23 November 2025
- Published by MDPI AG in Mathematics
- Vol. 13 (23) , 3753
Open AccessImproving the Performance of Static Malware Classification Using Deep Learning Models and Feature Reduction Strategies
The rapid evolution of malware continues to pose severe challenges to cybersecurity, highlighting the need for accurate and efficient detection systems. Traditional signature- and heuristic-based methods are increasingly inadequate against sophisticated threats, which has motivated the use of machine learning and deep learning for static malware classification. In this study, we propose three deep neural network (DNN) architectures tailored for the binary classification of Portable Executable (PE) files. The models were trained and validated on the EMBER 2017 dataset and further tested on the independent REWEMA dataset to evaluate their cross-dataset generalization capabilities. To address the computational burden of high-dimensional feature vectors, two feature reduction strategies were examined: the Kumar method, which selected 276 features, and the LightGBM-based intersection method, which identified 206 shared features. Experimental results showed that the proposed Model III consistently achieved the best overall performance, outperforming LightGBM (v3.3.5) and the other DNN models in terms of accuracy, recall, and F1-score. Notably, its recall exceeded that of LightGBM by 0.73%, highlighting its superiority in reducing false negative rates. Feature reduction further demonstrated that significant dimensionality reduction could be achieved without compromising classification quality, with the Kumar method achieving the best balance between accuracy and efficiency. Cross-dataset validation revealed performance degradation across all models due to distributional shifts, but the decline was less significant for the DNNs, confirming its greater adaptability compared with LightGBM. These findings demonstrate that architectural optimization and appropriate feature selection can significantly improve the performance of static malware classification. This study also provides empirical benchmarks and methodological guidance for developing accurate, efficient, and resilient malware detection systems that are resilient to evolving threats.
- Open Access
- 7 November 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (22) , 11862
Legacy Code, Live Risk: Empirical Evidence of Malware Detection Gaps
Consistent detection of malicious loaders across varied programming languages and build tools remains a significant cybersecurity challenge. This study empirically measures how compiler and language choices affect the detectability of standard in-memory Windows loaders. We implement functionally equivalent loaders (allocate, copy, protect, execute) in C, C#, Fortran, and COBOL, embedding an identical x64 test payload to isolate behavior. Our results reveal significant detection gaps: loaders compiled in legacy languages (Fortran, COBOL) consistently evade static and dynamic antivirus engines that easily flag their C and C# counterparts. We demonstrate this evasion is not due to behavioral differences, but to compiler-specific static artifacts. These artifacts, such as interleaved zero-bytes in Fortran and fragmented payload-construction logic in COBOL, effectively break common signature matching. These findings indicate that many detection tools are overly sensitive to the static build surface rather than true semantic behavior. We provide actionable guidance favoring behavior-focused analysis, such as tracking API call order and memory protection changes, to address this critical legacy code blind spot.
- 7 November 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (22) , 11862
Open AccessLegacy Code, Live Risk: Empirical Evidence of Malware Detection Gaps
Consistent detection of malicious loaders across varied programming languages and build tools remains a significant cybersecurity challenge. This study empirically measures how compiler and language choices affect the detectability of standard in-memory Windows loaders. We implement functionally equivalent loaders (allocate, copy, protect, execute) in C, C#, Fortran, and COBOL, embedding an identical x64 test payload to isolate behavior. Our results reveal significant detection gaps: loaders compiled in legacy languages (Fortran, COBOL) consistently evade static and dynamic antivirus engines that easily flag their C and C# counterparts. We demonstrate this evasion is not due to behavioral differences, but to compiler-specific static artifacts. These artifacts, such as interleaved zero-bytes in Fortran and fragmented payload-construction logic in COBOL, effectively break common signature matching. These findings indicate that many detection tools are overly sensitive to the static build surface rather than true semantic behavior. We provide actionable guidance favoring behavior-focused analysis, such as tracking API call order and memory protection changes, to address this critical legacy code blind spot.
- Open Access
Evading Antivirus Detection Using Fountain Code-Based Techniques for Executing Shellcodes
In this study, we propose a method for successfully evading antivirus detection by encoding malicious shellcode with fountain codes. The Meterpreter framework for Microsoft Windows 32-bit and 64-bit architectures was used to produce the shellcode used in this investigation. The experimental results proved that detection rates were substantially decreased. Specifically, the number of detected instances using antivirus vendors for 32-bit shellcode decreased from 18 to 3, while for 64-bit shellcode, it decreased from 16 to 1. This method breaks up a malicious payload into many packets, each with their own distinct structure, and then encodes them. This obfuscation approach maintains the shellcode’s integrity, ensuring correct code execution. However, in the persistence phase of the penetration testing process, this method offers an additional means of evading antivirus techniques.
Open AccessEvading Antivirus Detection Using Fountain Code-Based Techniques for Executing Shellcodes
In this study, we propose a method for successfully evading antivirus detection by encoding malicious shellcode with fountain codes. The Meterpreter framework for Microsoft Windows 32-bit and 64-bit architectures was used to produce the shellcode used in this investigation. The experimental results proved that detection rates were substantially decreased. Specifically, the number of detected instances using antivirus vendors for 32-bit shellcode decreased from 18 to 3, while for 64-bit shellcode, it decreased from 16 to 1. This method breaks up a malicious payload into many packets, each with their own distinct structure, and then encodes them. This obfuscation approach maintains the shellcode’s integrity, ensuring correct code execution. However, in the persistence phase of the penetration testing process, this method offers an additional means of evading antivirus techniques.
- Open Access
Chaotic-Based Shellcode Encryption: A New Strategy for Bypassing Antivirus Mechanisms
This study employed chaotic systems as an innovative approach for shellcode obfuscation to evade current antivirus detection methods. Standard AV solutions primarily rely on static signatures and heuristic analysis to identify malicious code. However, chaotic systems employ dynamic and unpredictable encryption methods, significantly obstructing detection efforts. The utilization of various chaotic maps for shellcode encryption facilitates the generation of multiple unique variations from the same functional code, each exhibiting distinct unpredictability due to the inherent nonlinearity and sensitivity of chaotic systems to initial conditions. The unpredictability of these situations poses a considerable challenge for antivirus software in recognizing consistent patterns, resulting in decreased detection rates. The findings from our experiments demonstrate that chaos-driven encryption methods significantly outperform traditional encryption techniques in terms of evading detection. This paper emphasizes the potential of chaos theory to enhance malware evasion strategies, offering a sophisticated approach to bypassing modern antivirus protections while ensuring the effectiveness of malicious payloads.
Open AccessChaotic-Based Shellcode Encryption: A New Strategy for Bypassing Antivirus Mechanisms
This study employed chaotic systems as an innovative approach for shellcode obfuscation to evade current antivirus detection methods. Standard AV solutions primarily rely on static signatures and heuristic analysis to identify malicious code. However, chaotic systems employ dynamic and unpredictable encryption methods, significantly obstructing detection efforts. The utilization of various chaotic maps for shellcode encryption facilitates the generation of multiple unique variations from the same functional code, each exhibiting distinct unpredictability due to the inherent nonlinearity and sensitivity of chaotic systems to initial conditions. The unpredictability of these situations poses a considerable challenge for antivirus software in recognizing consistent patterns, resulting in decreased detection rates. The findings from our experiments demonstrate that chaos-driven encryption methods significantly outperform traditional encryption techniques in terms of evading detection. This paper emphasizes the potential of chaos theory to enhance malware evasion strategies, offering a sophisticated approach to bypassing modern antivirus protections while ensuring the effectiveness of malicious payloads.
- Open Access
- 26 September 2021
- Published by Springer Nature in EURASIP Journal on Advances in Signal Processing
- Vol. 2021 (1) , 1-18
Statistical feature-based steganalysis for pixel-value differencing steganography
Pixel-value differencing (PVD) steganography is a popular spatial domain technology. Several PVD-based studies have proposed extended PVD steganography methods. The majority of these studies have verified their security against the regular-singular (RS) analysis. However, RS analysis is aimed at the feature of the least significant bit substitution method, which is relatively less significant for PVD steganography. The pixel difference histogram (PDH) is generally utilized to attack PVD steganography. If the embedding capacity is high, then the features on the PDH are evident; otherwise, the features are less obvious. In this paper, we propose a statistical feature-based steganalysis technique for the original PVD steganography. Experimental results demonstrate that, compared with existing steganalysis technique with weighted stego-image (WS) method, the proposed method effectively detects PVD steganography at low embedding ratios, such that there is no need of using the original embedding parameters. Furthermore, the accuracy and precision of the method are better than those of existing PVD steganalysis techniques. Therefore, the proposed method contributes to the security analysis of the original PVD steganography as an alternative to the commonly used RS, PDH and WS attack techniques.
- 26 September 2021
- Published by Springer Nature in EURASIP Journal on Advances in Signal Processing
- Vol. 2021 (1) , 1-18
Open AccessStatistical feature-based steganalysis for pixel-value differencing steganography
Pixel-value differencing (PVD) steganography is a popular spatial domain technology. Several PVD-based studies have proposed extended PVD steganography methods. The majority of these studies have verified their security against the regular-singular (RS) analysis. However, RS analysis is aimed at the feature of the least significant bit substitution method, which is relatively less significant for PVD steganography. The pixel difference histogram (PDH) is generally utilized to attack PVD steganography. If the embedding capacity is high, then the features on the PDH are evident; otherwise, the features are less obvious. In this paper, we propose a statistical feature-based steganalysis technique for the original PVD steganography. Experimental results demonstrate that, compared with existing steganalysis technique with weighted stego-image (WS) method, the proposed method effectively detects PVD steganography at low embedding ratios, such that there is no need of using the original embedding parameters. Furthermore, the accuracy and precision of the method are better than those of existing PVD steganalysis techniques. Therefore, the proposed method contributes to the security analysis of the original PVD steganography as an alternative to the commonly used RS, PDH and WS attack techniques.
- 2 April 2021
- Published by Springer Nature in Arabian Journal for Science and Engineering
- Vol. 46 (9) , 8525-8533
Chi-square-based steganalysis method against modified pixel-value differencing steganography
- 2 April 2021
- Published by Springer Nature in Arabian Journal for Science and Engineering
- Vol. 46 (9) , 8525-8533
Chi-square-based steganalysis method against modified pixel-value differencing steganography