No abstract

We report a novel demodulation method that enables single snapshot wide field imaging of optical properties of turbid media in the Spatial Frequency Domain (SFD). This Single Snapshot Multiple frequency Demodulation (SSMD) method makes use of the orthogonality of harmonic functions to extract the modulation transfer function (MTF) at multiple modulation frequencies simultaneously from a single structured-illuminated image at once. The orientation, frequency, and amplitude of each modulation can be set arbitrarily subject to the limitation of the implementation device. We first validate and compare SSMD to the existing demodulation methods by numerical simulations. The performance of SSMD is then demonstrated with experiments on both tissue mimicking phantoms and in vivo for recovering optical properties by comparing to the standard three-phase demodulation approach. The results show that SSMD increases significantly the data acquisition speed and reduces motion artefacts. SSMD exhibits excellent noise suppression in imaging as well at the rate proportional to the square root of the number of pixels contained in its kernel. SSMD is ideal in the implementation of a real-time spatial frequency domain imaging platform and will open up SFDI for vast applications in imaging and monitoring dynamic turbid medium and processes.

Deep Neural Network (DNN) has been showing great potential in kinds of real-world applications such as fraud detection and distress prediction. Meanwhile, data isolation has become a serious problem currently, i.e., different parties cannot share data with each other. To solve this issue, most research leverages cryptographic techniques to train secure DNN models for multi-parties without compromising their private data. Although such methods have strong security guarantee, they are difficult to scale to deep networks and large datasets due to its high communication and computation complexities. To solve the scalability of the existing secure Deep Neural Network (DNN) in data isolation scenarios, in this paper, we propose an industrial scale privacy preserving neural network learning paradigm, which is secure against semi-honest adversaries. Our main idea is to split the computation graph of DNN into two parts, i.e., the computations related to private data are performed by each party using cryptographic techniques, and the rest computations are done by a neutral server with high computation ability. We also present a defender mechanism for further privacy protection. We conduct experiments on real-world fraud detection dataset and financial distress prediction dataset, the encouraging results demonstrate the practicalness of our proposal.

No abstract

No abstract