Nov 2, 2018
Loptata Hall, Room 101
Autoencoder Embedding of Task-Specific Information
Adviser: Mark Anastasio
Task-based measures of image quality (IQ) quantify the ability of an observer to perform a specific task. Such measures are employed for assessing and optimizing medical imaging systems. Although the Bayesian ideal observer is optimal by definition, it is frequently both non-linear and intractable. In such cases, linear observers are commonly employed. However, the optimal linear observer, the Hotelling observer (HO), becomes intractable when considering large images. Channelized methods have become popular for reducing the dimensionality of image data, but they lack a consistent design process. Autoencoders (AEs) are neural networks that are used to learn concise representations of data, frequently for the purposes of reducing dimensionality. In this work, we investigate the amount of task specific information lost when applying an autoencoder to reduce the dimensionality of data before performing a detection task. We trained several autoencoders to encode task-specific information by modifying the standard loss function and examined the effect of hidden layer size and the use of tied/untied weights on the resulting representation accuracy. Subsequently, HOs were applied to both the original images and the dimensionality-reduced versions of them produced by the AEs. It was demonstrated that, for a suitable specification of the AE, the performance of the HO was relatively unaffected by the encoding of the image. However, the computational cost of inverting the covariance matrix was greatly reduced when the HO was applied with the encoded data due to its reduced dimensionality. Our findings suggest that AEs may represent an attractive alternative to the use of heuristic channels for reducing the dimensionality of image data when seeking to accurately approximate the performance of the HO on signal detection tasks.
Efficient Holistic Control over Industrial Wireless Sensor-Actuator Networks
Adviser: Chenyang Lu
Process automation is embracing wireless sensor-actuator networks (WSANs) in the era of Industrial Internet. Despite the success of WSANs for monitoring applications, feedback control poses significant challenges due to data loss and stringent energy constraints in WSANs. Holistic control adopts a cyber-physical system approach to overcome the challenges by orchestrating network reconfiguration and process control at run time. We explore efficient holistic control designs to maintain control performance while reducing the communication cost. There are four parts in this talk: (1) introduce a holistic control architecture that integrates low-power wireless bus (LWB) and two control strategies, rate adaptation and self-triggered control, specifically proposed to reduce communication cost; (2) propose a novel wireless network mechanisms to support rate adaptation and self-triggered control, respectively, in a multi-hop WSAN; (3) present a real-time network-in-the-loop simulator that integrates MATLAB/Simulink and a three-floor WSAN testbed to evaluate wireless control systems; (4) explore the tradeoff between communication cost and control performance under alternative holistic control approaches in variable cyber and physical conditions.