Biblio

What makes a security visualization effective? How do we measure visualization effectiveness in the context of investigating, analyzing, understanding and reporting cyber security incidents? Identifying and understanding cyber-attacks are critical for decision making - not just at the technical level, but also the management and policy-making levels. Our research studied both questions and extends our Security Visualization Effectiveness Measurement (SvEm) framework by providing a full-scale effectiveness approach for both theoretical and user-centric visualization techniques. Our framework facilitates effectiveness through interactive three-dimensional visualization to enhance both single and multi-user collaboration. We investigated effectiveness metrics including (1) visual clarity, (2) visibility, (3) distortion rates and (4) user response (viewing) times. The SvEm framework key components are: (1) mobile display dimension and resolution factor, (2) security incident entities, (3) user cognition activators and alerts, (4) threat scoring system, (5) working memory load and (6) color usage management. To evaluate our full-scale security visualization effectiveness framework, we developed VisualProgger - a real-time security visualization application (web and mobile) visualizing data provenance changes in SvEm use cases. Finally, the SvEm visualizations aims to gain the users' attention span by ensuring a consistency in the viewer's cognitive load, while increasing the viewer's working memory load. In return, users have high potential to gain security insights in security visualization. Our evaluation shows that viewers perform better with prior knowledge (working memory load) of security events and that circular visualization designs attract and maintain the viewer's attention span. These discoveries revealed research directions for future work relating to measurement of security visualization effectiveness.

We construct targeted audio adversarial examples on automatic speech recognition. Given any audio waveform, we can produce another that is over 99.9% similar, but transcribes as any phrase we choose (recognizing up to 50 characters per second of audio). We apply our white-box iterative optimization-based attack to Mozilla's implementation DeepSpeech end-to-end, and show it has a 100% success rate. The feasibility of this attack introduce a new domain to study adversarial examples.

Compressed sensing can represent the sparse signal with a small number of measurements compared to Nyquist-rate samples. Considering the high-complexity of reconstruction algorithms in CS, recently compressive detection is proposed, which performs detection directly in compressive domain without reconstruction. Different from existing work that generally considers the measurements corrupted by dense noises, this paper studies the compressive detection problem when the measurements are corrupted by both dense noises and sparse errors. The sparse errors exist in many practical systems, such as the ones affected by impulse noise or narrowband interference. We derive the theoretical performance of compressive detection when the sparse error is either deterministic or random. The theoretical results are further verified by simulations.

Most Depth Image Based Rendering (DIBR) techniques produce synthesized images which contain non-uniform geometric distortions affecting edges coherency. This type of distortions are challenging for common image quality metrics. Morphological filters maintain important geometric information such as edges across different resolution levels. There is inherent congruence between the morphological pyramid decomposition scheme and human visual perception. In this paper, multi-scale measure, morphological pyramid peak signal-to-noise ratio MP-PSNR, based on morphological pyramid decomposition is proposed for the evaluation of DIBR synthesized images. It is shown that MPPSNR achieves much higher correlation with human judgment compared to the state-of-the-art image quality measures in this context.