PrefaceDL was developed in our research group over the past 15, or so, years. The book disseminates this breakthrough mathematical-engineering idea, which results in 100 times improvement and better in classical algorithmic areas that have been intensively studied for decades. Initial developments in DL were described in "Neural Networks and Intellect," by L. Perlovsky, Oxford University Press, 2001 (which is now in the 3rd printing). The current book describes new breakthrough results developed during the last eight years. First we present the basic technique of DL, explain the fundamental mathematical reason why classical techniques in many areas fail for real-world problems, and how DL overcomes this difficulty. We discuss the algorithmic failure of many techniques to reach informationtheoretic performance bounds, relate it to computational complexity, and ultimately to the Gödel theory (it turns out that all past algorithms, neural networks, fuzzy systems, used logic at some step and were subject to Gödelian limitations).Then we describe a number of applications where significant breakthrough improvements were achieved over popular state-of-the-art techniques (detection, clustering, supervised and unsupervised learning, tracking, sensor fusion, prediction, and particularly financial prediction). We follow with novel engineering areas, where revolutionary results were obtained. The theory is extended toward mathematical modeling of the mind, including higher cognitive functions, beyond anything that has been published in engineering books (no competition): mechanisms of the mind-brain (recent neuroimaging experiments proved that brain is actually using DL computations), applications to learning natural language, to language-understanding search engines for the Internet, to modeling interactions between language and cognition, language and emotions, evolution of languages, evolution of cultures, the role of music in evolution of the mind and cultures.The mind is the best mechanism for solving complex engineering problems. Therefore, it is just natural that developing engineering algorithms and modeling the mind goes hand in hand. Solving complex engineering problems helps understand working of the mind, and cognitively-inspired algorithms work better than classical engineering methods. This approach to engineering is called computational intelligence.The book is based on about 200 papers published over the last several years describing DL and its applications. Many of them were important events attracting attention and receiving awards. Every book chapter is written anew, all are unified by a common theme -mathematical technique of dynamic logic and by consistent notations. The book is written for students as well as seasoned professionals, it VI Preface contains details about applications, algorithms, notations, flowcharts, details that are missing in the papers. DL is easy to use as a textbook or manual. Engineering improvements achieved make it stand out over other texts.The book contains two parallel tracks...

A generalized diffraction tomographic (DT) algorithm is derived for subsurface imaging from multifrequency multi-monostatic ground penetrating radar (GPR) data. The algorithm is based on the Born approximation for vector electromagnetic scattering and incorporates realistic nearfield models for the receiving and transmitting antennas. The forward scattering model is inverted analytically using the regularized pseudoinverse operator to yield an algorithm for imaging the underground region based on scattered field measurements at a set of receiving antennas. Whereas the usual inversion algorithms of DT require a lossless background medium and ideal point sources and receivers, the algorithm described here allows an attenuating background and arbitrary transmitting and receiving antennas. The algorithm places no restrictions on the radar frequency, and can thus include shallow imaging applications where the wavelengths are on the same order as the depth of buried objects of interest. Versions of the algorithm are given for both the three dimensional and the 2.5-dimensional cases. Results are given of computer simulations designed to test the algorithm.

No abstract

No abstract

A new approach is described for combining range and Doppler data from multiple radar platforms to perform multi-target detection and tracking. In particular, azimuthal measurements are assumed to be either coarse or unavailable, so that multiple sensors are required to triangulate target tracks using range and Doppler measurements only. Increasing the number of sensors can cause data association by conventional means to become impractical due to combinatorial complexity, i.e., an exponential increase in the number of mappings between signatures and target models. When the azimuthal resolution is coarse, this problem will be exacerbated by the resulting overlap between signatures from multiple targets and clutter. In the new approach, the data association is performed probabilistically, using a variation of expectation-maximization (EM). Combinatorial complexity is avoided by performing an efficient optimization in the space of all target tracks and mappings between tracks and data. The full, multi-sensor, version of the algorithm is tested on simulated data. The results demonstrate that accurate tracks can be estimated by exploiting spatial diversity in the sensor locations. Also, as a proof-of-concept, a simplified, single-sensor range-only version of the algorithm is tested on experimental radar data acquired with a stretch radar receiver. These results are promising, and demonstrate robustness in the presence of nonhomogeneous clutter.