Charles’ thesis, entitled “Dynamics and Correlations in Sparse Signal Acquisition,” centers on developing algorithms to analyze time-varying high-dimensional data by exploiting the fact that the information of interest is generally low-dimensional. His work specifically advances algorithms for a particular kind of structure called "sparsity" that has achieved state-of-the-art results in many signal/image processing algorithms. As one example, Charles used his algorithms on a remote sensing task to show that he could take common multispectral imaging and improve the spectral resolution by an order of magnitude to a quality comparable to data from a (much more expensive) hyperspectral imager. In addition to engineering applications, Charles’ work has also provided fundamental analysis of neural network structures for working memory that gives insight into the performance of both biological and machine learning systems. Advised by ECE Associate Professor Christopher J. Rozell, Charles graduated in May 2015 and is now a postdoctoral fellow at Princeton University.

Sodagar’s thesis, entitled “Enabling Integrated Nanophotonic Devices in Hybrid CMOS-Compatible Material Platforms for Optical Interconnection,” mainly focused on the design and development of a high-speed electro-optic modulator for silicon photonic platforms. Sodagar also demonstrated a series of essential integrated photonic elements that are urgently needed for realization of complex integrated photonic systems. The outcome of his research impacts the telecommunication industry. As an example, his work can be adopted for the next generation of high-speed links within datacenters. Advised by ECE Professor Ali Adibi, Sodagar graduated in August 2015 and is now a senior engineer at Skorpios Technologies, Inc.

Trivedi’s thesis, entitled “Ultra-Low Power Non-Boolean Computing with Tunneling Field-Effect Transistors,” demonstrated evolutionary and disruptive approaches of integrating Tunneling Field-Effect-Transistor (Tunnel FET) technology with non-Boolean computing. In the evolutionary approach, silicon channel Tunnel FETs were applied to Cellular Neural Networks (CNNs) for energy-efficient image processing and associative memory. In the disruptive approach, an atypical gate/source-overlapped heterojunction Tunnel FET (SO-HTFET) was demonstrated to exhibit Gaussian-shaped I_DS-V_GS. The design approaches discussed in Trivedi’s thesis will be useful to enhance computing abilities of low power computing platforms. As examples, medical diagnosis in wearable healthcare devices can be improved and remote sensors can better assess their environment. Advised by ECE Associate Professor Saibal Mukhopadhyay, Trivedi graduated in Fall 2015 and is now an assistant professor in the Department of Electrical and Computer Engineering at the University of Illinois at Chicago.