There is, I think, something ethereal about i —the square root of minus one. I remember first hearing about it at school. It seemed an odd beast at that time—an intruder hovering on the edge of reality.

Usually familiarity dulls this sense of the bizarre, but in the case of i it was the reverse: over the years the sense of its surreal nature intensified. It seemed that it was impossible to write mathematics that described the real world in …

I and i

다중혈관 관상동맥 환자에서 y-문합을 이용하여 양쪽 내흉동맥만을 사용한 우회술의 조기 성적

Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

https://minerva-access.unimelb.edu.au/bitstream/handle/11343/197439/1207.0203v1.pdf

Internet of Things (IoT): A vision, architectural elements, and future directions

We propose a deep learning method for single image super-resolution (SR). Our method directly learns an end-to-end mapping between the low/high-resolution images. The mapping is represented as a deep convolutional neural network (CNN) that takes the low-resolution image as the input and outputs the high-resolution one. We further show that traditional sparse-coding-based SR methods can also be viewed as a deep convolutional network. But unlike traditional methods that handle each component separately, our method jointly optimizes all layers. Our deep CNN has a lightweight structure, yet demonstrates state-of-the-art restoration quality, and achieves fast speed for practical on-line usage. We explore different network structures and parameter settings to achieve trade-offs between performance and speed. Moreover, we extend our network to cope with three color channels simultaneously, and show better overall reconstruction quality.

Image Super-Resolution Using Deep Convolutional Networks

Macro x-ray ﬂuorescence (XRF) imaging of cultural heritage objects, while a popular non-invasive technique for providing elemental distribution maps, is a slow acquisition process in acquiring high signal-to-noise ratio XRF volumes. Typically on the order of tenths of a second per pixel, a raster scanning probe counts the number of photons at different energies emitted by the object under x-ray illumination. In an effort to reduce the scan times without sacriﬁcing elemental map and XRF volume quality, we propose using dictionary learning with a Poisson noise model as well as a color image-based prior to restore noisy, rapidly acquired XRF data.

Denoising Fast X-Ray Fluorescence Raster Scans of Paintings

Disclosed is an image encoder that divides (1000) a digital image into a set of "macroblocks." If appropriate, a macroblock is "downsampled" (1004) to a lower resolution. The lower-resolution macroblock is then encoded by applying spatial (and possibly temporal) prediction (1006). The "residual" of the macroblock is calculated (1010) as the difference between the predicted and actual contents of the macroblock. The low-resolution residual is then either transmitted to an image decoder or stored for later use (1010). In some embodiments, the encoder calculates (1008) the rate-distortion costs of encoding the original-resolution macroblock and the lower-resolution macroblock and then only encodes (1010) the lower-resolution macroblock if its cost is lower. When a decoder receives (1104) a lower-resolution residual, it recovers the lower-resolution macroblock using standard prediction techniques (1106). Then, the macroblock is "upsampled" (1110) to its original resolution by interpolating the values left out by the encoder. The macroblocks are then joined (1114) to form the original digital image.

Digital image compression by adaptive macroblock resolution coding

While Convolutional Neural Networks (CNNs) trained for image and video super-resolution (SR) regularly achieve new state-of-the-art performance, they also suffer from significant drawbacks. One of their limitations is their lack of robustness to unseen image formation models during training. Other limitations include the generation of artifacts and hallucinated content when training Generative Adversarial Networks (GANs) for SR. While the Deep Learning literature focuses on presenting new training schemes and settings to resolve these various issues, we show that one can avoid training and correct for SR results with a fully self-supervised fine-tuning approach. More specifically, at test time, given an image and its known image formation model, we fine-tune the parameters of the trained network and iteratively update them using a data fidelity loss. We apply our fine-tuning algorithm on multiple image and video SR CNNs and show that it can successfully correct for a sub-optimal SR solution by entirely relying on internal learning at test time. We apply our method on the problem of fine-tuning for unseen image formation models and on removal of artifacts introduced by GANs.

Self-supervised Fine-tuning for Correcting Super-Resolution Convolutional Neural Networks

In this paper, a distributed detection approach for displacement estimation in image sequences is presented. This method is derived from a Bayesian framework and reduces to a M-ary Hypothesis test among a representative set of possible displacement vectors. It is shown that the mean-squared error based block-matching (BM) algorithm is a special case of this general approach. In our approach, at each point of the current frame a set of overlapping localized detectors outputs a number of estimates for the displacement vector. Then, a distributed detection network is adopted for the fusion of the these estimates. Since the computational load is high, suboptimal but computationally efficient solutions are proposed. The above method gives a more accurate estimation of the displacement field and it is shown to be more robust in the presence of occlusion and noise, compared to the BM algorithm. Experimental results on video-conference image sequences are presented.

Distributed detection methods for displacement estimation

We solve the problem of uplink video streaming in CDMA cellular networks by jointly designing the rate control and scheduling algorithms. In the pricing-based distributed rate control algorithm, the base station announces a price for the per unit average rate it can support, and the mobile devices choose their desired average transmission rates by balancing their video quality and cost of transmission. Each mobile device then determines the specific video frames to transmit by a video summarization process. In the time-division-multiplexing (TDM) scheduling algorithm, the base station collects the information on frames to be transmitted from all devices within the current time window, sorts them in increasing order of deadlines, and schedules the transmissions in a TDM fashion. This joint algorithm takes advantage of the multi-user content diversity, and maximizes the network total utility (i.e., minimize the network total distortion), while satisfying the delivery deadline constraints. Simulations showed that the proposed algorithm significantly outperforms the constant rate provision algorithm.

/pdf/joint-rate-control-and-scheduling-for-wireless-uplink-video-3qwf1q0tan.pdf

Aggelos K. Katsaggelos

Papers

Denoising Fast X-Ray Fluorescence Raster Scans of Paintings

Digital image compression by adaptive macroblock resolution coding

Self-supervised Fine-tuning for Correcting Super-Resolution Convolutional Neural Networks

Distributed detection methods for displacement estimation

Joint rate control and scheduling for wireless uplink video streaming