PhD Alumni
Dr. Felix Anda
University College Dublin
Felix Anda completed his PhD in the School of Computer Science, UCD, under the co-supervision of Dr. Mark Scanlon and Dr. Nhien-An Le-Khac.
His PhD research focused on automated, machine learning-based digital evidence classification and identification techniques.
Research Output
Publications
Digital Forensics: Leveraging Deep Learning Techniques in Facial Images to Assist Cybercrime Investigations
School of Computer Science, University College Dublin
This PhD thesis presents a novel approach to facial age estimation using deep learning techniques to assist cybercrime investigations. The research addresses the digital forensic backlog by proposing age estimation models that surpass the state-of-the-art facial age detectors for subjects under 25. The study evaluates the performance of various image pre-processing techniques, neural network architectures, and hyper-parameter optimisation strategies.
Vec2UAge: Enhancing Underage Age Estimation Performance through Facial Embeddings
Forensic Science International: Digital Investigation
This paper presents Vec2UAge, a novel regression-based model for estimating the age of underage individuals from facial embeddings. The model is trained on the VisAGe and Selfie-FV datasets and achieves a mean absolute error rate of 2.36 years. The authors evaluate the impact of random initializations, optimizers, and learning rates on the model's performance.
SoK: Exploring the State of the Art and the Future Potential of Artificial Intelligence in Digital Forensic Investigation
The 13th International Workshop on Digital Forensics (WSDF), held at the 15th International Conference on Availability, Reliability and Security (ARES)
This systematic overview of artificial intelligence (AI) in digital forensic investigation explores the current state of the art and future potential of AI in expediting digital forensic analysis and increasing case processing capacities. The authors discuss AI applications in data discovery, device triage, and other areas, highlighting current challenges and future directions.
Assessing the Influencing Factors on the Accuracy of Underage Facial Age Estimation
The 6th IEEE International Conference on Cyber Security and Protection of Digital Services (Cyber Security)
This study evaluates the influencing factors on the accuracy of underage facial age estimation using two cloud services, Microsoft Azure's Face API and Amazon Web Service's Rekognition service. The analysis of the VisAGe dataset reveals correlations between facial attributes and age estimation errors, identifying the most significant factors to be addressed in future age estimation modeling.
DeepUAge: Improving Underage Age Estimation Accuracy to Aid CSEM Investigation
Forensic Science International: Digital Investigation Vol. 32 pp. 300921
DeepUAge improves underage age estimation accuracy to aid Child Sexual Exploitation Material (CSEM) investigation. The model, trained on the VisAGe dataset, achieves state-of-the-art performance for age estimation of minors, with a mean absolute error (MAE) rate of 2.73 years. This work tackles the challenges of collecting and annotating underage facial age data, and its application can expedite digital investigations.
Improving Borderline Adulthood Facial Age Estimation through Ensemble Learning
The 8th International Workshop on Cyber Crime (IWCC), held at the 14th International Conference on Availability, Reliability and Security (ARES)
This paper presents an ensemble learning approach to improve facial age estimation for borderline adulthood cases. The authors develop a deep learning model (DS13K) and fine-tune it on the Deep Expectation (DEX) model to achieve an accuracy of 68% for the age group 16-17 years old, outperforming DEX by 4 times. The study also evaluates existing cloud-based facial age prediction services.
Improving the Accuracy of Automated Facial Age Estimation to Aid CSEM Investigations
Digital Investigation Vol. 28 pp. S142
This study evaluates existing age prediction services and introduces a deep learning model, DS13K, to improve the accuracy of underage facial age estimation in child sexual exploitation material (CSEM) investigations. The model outperforms existing services, particularly in the borderline adulthood age range (16-17 years old), with an accuracy rate of 68%.
Evaluating Automated Facial Age Estimation Techniques for Digital Forensics
12th International Workshop on Systematic Approaches to Digital Forensics Engineering (SADFE), IEEE Security & Privacy Workshops
This paper evaluates existing automated facial age estimation techniques for digital forensics, highlighting their limitations and proposing a dataset generator to overcome the lack of sufficient sample images in specific age ranges. The study assesses the performance of offline and cloud-based models, releasing a tool to generate uniformly distributed random images by age and gender.