Software dependability and security are critical in assuring the resilience of these complex systems. Despite decades of work in this area, software remains a weak link in system integrity, leading to failures that compromise safety and/or impose financial costs. The challenge posed is at once of critical importance and immense. We believe progress is best made through a new approach that focuses on mitigating the types of software bugs that are most difficult to address with conventional methods, and the team we have assembled is singularly well qualified to pursue this path. To meet the challenge, the proposed program will carry out education in the area of software faults, failures and their mitigations at development cycle and specifically during system operation. We have pioneered a course for graduate students, young researchers and software engineers studying or working in software engineering field. This new course is playing an important role in both the masters program for electrical and computer engineering and the undergraduate program for interdisciplinary data science at Duke Kunshan University.
Modern life depends on devices and systems containing a moderate to significant amount of software whose reliability is critical to the reliability of a system as a whole. Software fault-tolerance has hitherto been based on design diversity, and its high implementation cost has largely limited the scope of application to safety-critical systems. Affordable software fault tolerance using the newer notion of environmental diversity is being studied in this project. The key idea is predicated on the existence of elusive software faults known as environment dependent bugs or Mandelbugs with transient characteristics in their manifestation. The environment for a software system here is taken to mean the operating systems resources and other concurrently running applications. This project mainly focuses on the following four research aspects: environmental factor identification, key environmental control techniques, environmentally diversity-based fault tolerance approaches, and applications to Android systems. Research on the failure data analysis, experimental research with accelerated life testing, analytic modeling and optimization techniques of open source software is being carried out. The fruits of the research will effectively contribute to reduction of the cost of software fault tolerance, while reducing the impact of environment dependent bugs on software reliability/availability. It will also contribute to the emergence and development of the environment dependent bugs related research in software engineering.
Prof. Ming Li and his lab conduct research in the area of Audio, Speech and Language Processing. In the 2020 calendar year, they have published more than 10 top conference or journal papers in this filed. The topics include speaker recognition, speaker diarization, speech synthesis, spoken term detection, paralinguistic speech attribute recognition, etc. They have collaborated with multiple industry leaders and local companies in terms of collaborative research and technology transfer.
Prof. Ming Li and his lab conduct research in the area of Multimodal Behavior Signal Analysis and Interpretation towards the AI assisted Autism Spectrum Disorder (ASD) diagnose. They have developed an AI studio for the early screening of ASD. The studio’s four walls are programmable projection screens that can recreate a variety of settings, such as a forest environment, with sound delivered through multichannel audio equipment. The therapist can use the studio to interact with the child, such as asking him or her to point at a certain object projected onto the wall to observe their reaction. At the same time, cameras capture the movements of the child and the therapist, including gestures, gazes and other actions. The studio is equipped with more than 10 technologies that have obtained or are in the process of obtaining patents. These include technologies that assist with gaze detection, human pose estimation, face detection, face recognition; speech recognition and paralinguistic attribute detection.
In the research project, we have designed two efficient end-to-cloud collaboration intelligence framework. The first one is PCCNN, a Convolutional Neural Network (CNN) partitioning method. It first compresses a CNN to generate new layers that can serve as candidate partitioning points, then trains prediction models to find an optimal partitioning point and splits the compressed CNN model into two parts deployed on the terminal device and the cloud, respectively.
The second one is a hierarchical data filtering framework for distributed deep neural networks, called dfDDNNs, that can avoid unnecessary transmission and cloud computing costs. Based on depthwise separable convolutions, we design a lightweight data filtering module utilized to identify and filter out the data that the cloud cannot recognize. Extensive experimental results demonstrate that the accuracy of the designed data filtering module is up to 83.18% in identifying worthless data and the proposed hierarchical data filtering distributed framework can effectively save up to 63.07% of bandwidth.