We propose a deep convolutional neural network architecture codenamed Inception that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14). The main hallmark of this architecture is the improved utilization of the computing resources inside the network. By a carefully crafted design, we increased the depth and width of the network while keeping the computational budget constant. To optimize quality, the architectural decisions were based on the Hebbian principle and the intuition of multi-scale processing. One particular incarnation used in our submission for ILSVRC14 is called GoogLeNet, a 22 layers deep network, the quality of which is assessed in the context of classification and detection.

/pdf/going-deeper-with-convolutions-1yobw2o2ds.pdf

Going deeper with convolutions

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

Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

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

计量经济分析 = Econometric analysis

Forcing functions represent an important class of Boolean functions that have been extensively studied in the analysis of the dynamics of random Boolean networks as models of genetic regulatory systems. Several other so-called Post classes of Boolean functions are closely related to forcing functions and have been used in learning theory as well as in control systems. We develop novel spectral algorithms to test membership of a Boolean function in these classes. These algorithms are highly efficient and are essential in learning problems, especially in the context of genetic regulatory networks, where the same learning procedures are applied repeatedly.

Spectral methods for testing membership in certain post classes and the class of forcing functions

This paper presents a novel effective method for denoising of images corrupted by signal-dependent noise. Denoising is performed by coefficient shrinkage in the shape-adaptive DCT (SA-DCT) transform-domain. The Anisotropic Local Polynomial Approximation (LPA) — Intersection of Confidence Intervals (ICI) technique is used to define the shape of the transform's support in a pointwise adaptive manner. The use of such an adaptive transform support enables both a simpler modelling of the noise in the transform domain and a sparser decomposition of the signal. Consequently, coefficient shrinkage is very effective and the reconstructed estimate's quality is high, in terms of both numerical error-criteria and visual appearance, with sharp detail preservation and clean edges. Simulation experiments demonstrate the superior performance of the proposed algorithm for a wide class of noise models with a signal-dependent variance, including Poissonian (photon-limited imaging), film-grain, and speckle noise.

Signal-dependent noise removal in pointwise shape-adaptive DCT domain with locally adaptive variance

In this contribution, we propose a near-lossless compression algorithm for Color Filter Arrays (CFA) images. It allows higher compression ratio than any strictly lossless algorithm for the price of some small and controllable error. In our approach a structural transformation is applied first in order to pack the pixels of the same color in a structure appropriate for the subsequent compression algorithm. The transformed data is compressed by a modified version of the JPEG-LS algorithm. A nonlinear and adaptive error quantization function is embedded in the JPEG-LS algorithm after the fixed and context adaptive predictors. It is step-like and adapts to the base signal level in such a manner that higher error values are allowed for lighter parts with no visual qua lity loss. These higher error values are then suppressed by gamma correction applied during the image reconstruction stage. The algorithm can be adjusted for arbitrary pixel resolution, gamma value and allowable error range. The compression performance of the proposed algorithm has been tested for real CFA raw data. The results are presented in terms of compression ratio versus reconstruction error and the visual quality of the reconstructed images is demonstrated as well. Keywords: Bayer pattern, color filter array, near-lossless compression, JPEG-LS

Near-lossless compression algorithm for Bayer pattern color filter arrays

We investigate the applicability of modified B-spline functions in digital image resampling tasks. We consider a modification that is based on the use of a linear combination of shifted B-splines of different degrees. By optimizing the weighting coefficients, we get interpolating functions with improved frequency responses – flatter characteristics in the passband and a higher attenuation of unwanted images of the original baseband. We realize a computationally efficient procedure, based on the use of a modified Farrow structure. In our experiments, the modified B-spline decimators and interpolators have shown a better performance and a better visual quality than their classical B-splines counterparts.

Digital image resampling by modified b-spline functions

A general two-stage multiple description coding (MDC) scheme using whitening transform is analyzed. It represents the original image in a form of a coarse image approximation and a residual image. The coarse approximation is subsequently duplicated and combined with the residual image further split into two descriptions using a checkerboard block transform coefficients rearrangement. We identify the importance of a good coarse approximation and explore different approaches for changing its resolution and coding it. We also propose different approaches for coding the residual signal. The coder scheme is quite simple and yet achieves high performance comparable with other MDC methods.

Karen Egiazarian

Papers

Spectral methods for testing membership in certain post classes and the class of forcing functions

Signal-dependent noise removal in pointwise shape-adaptive DCT domain with locally adaptive variance

Near-lossless compression algorithm for Bayer pattern color filter arrays

Digital image resampling by modified b-spline functions

Two-stage multiple description image coders: Analysis and comparative study