A Connection Between GRBF and MLP

Both multilayer perceptrons (MLP) and Generalized Radial Basis Functions (GRBF) have good approximation properties, theoretically and experimentally. Are they related? The main point of this paper is to show that for normalized inputs, multilayer perceptron networks are radial function networks (albeit with a non-standard radial function). This provides an interpretation of the weights w as centers t of the radial function network, and therefore as equivalent to templates. This insight may be useful for practical applications, including better initialization procedures for MLP. In the remainder of the paper, we discuss the relation between the radial functions that correspond to the sigmoid for normalized inputs and well-behaved radial basis functions, such as the Gaussian. In particular, we observe that the radial function associated with the sigmoid is an activation function that is good approximation to Gaussian basis functions for a range of values of the bias parameter. The implication is that a MLP network can always simulate a Gaussian GRBF network (with the same number of units but less parameters); the converse is true only for certain values of the bias parameter. Numerical experiments indicate that this constraint is not always satisfied in practice by MLP networks trained with backpropagation. Multiscale GRBF networks, on the other hand, can approximate MLP networks with a similar number of parameters.

Face Recognition Under Varying Pose

While researchers in computer vision and pattern recognition have worked on automatic techniques for recognizing faces for the last 20 years, most systems specialize on frontal views of the face. We present a face recognizer that works under varying pose, the difficult part of which is to handle face rotations in depth. Building on successful template-based systems, our basic approach is to represent faces with templates from multiple model views that cover different poses from the viewing sphere. Our system has achieved a recognition rate of 98% on a data base of 62 people containing 10 testing and 15 modelling views per person.

Viewer-Centered Object Recognition in Monkeys

How does the brain recognize three-dimensional objects? We trained monkeys to recognize computer rendered objects presented from an arbitrarily chosen training view, and subsequently tested their ability to generalize recognition for other views. Our results provide additional evidence in favor of with a recognition model that accomplishes view-invariant performance by storing a limited number of object views or templates together with the capacity to interpolate between the templates (Poggio and Edelman, 1990).

Model-Based Matching by Linear Combinations of Prototypes

We describe a method for modeling object classes (such as faces) using 2D example images and an algorithm for matching a model to a novel image. The object class models are "learned'' from example images that we call prototypes. In addition to the images, the pixelwise correspondences between a reference prototype and each of the other prototypes must also be provided. Thus a model consists of a linear combination of prototypical shapes and textures. A stochastic gradient descent algorithm is used to match a model to a novel image by minimizing the error between the model and the novel image. Example models are shown as well as example matches to novel images. The robustness of the matching algorithm is also evaluated. The technique can be used for a number of applications including the computation of correspondence between novel images of a certain known class, object recognition, image synthesis and image compression.

Fast Object Recognition in Noisy Images Using Simulated Annealing

A fast simulated annealing algorithm is developed for automatic object recognition. The normalized correlation coefficient is used as a measure of the match between a hypothesized object and an image. Templates are generated on-line during the search by transforming model images. Simulated annealing reduces the search time by orders of magnitude with respect to an exhaustive search. The algorithm is applied to the problem of how landmarks, for example, traffic signs, can be recognized by an autonomous vehicle or a navigating robot. The algorithm works well in noisy, real-world images of complicated scenes for model images with high information content.

Component based recognition of objects in an office environment

We present a component-based approach for recognizing objects under large pose changes. From a set of training images of a given object we extract a large number of components which are clustered based on the similarity of their image features and their locations within the object image. The cluster centers build an initial set of component templates from which we select a subset for the final recognizer. In experiments we evaluate different sizes and types of components and three standard techniques for component selection. The component classifiers are finally compared to global classifiers on a database of four objects.

Carbon Nanotube Growth Using Ni Catalyst in Different Layouts

Vertically aligned carbon nanotubes have been grown using Ni as catalyst by plasma enhanced chemical vapor deposition system (PECVD) in various pre-patterned substrates. Ni was thermally evaporated on silicon substrates with anodized alumina mask prepared in different methods including 2 step anodization of porous alumina template and interference lithography assisted array of pores. The templates helped to define Ni nanodots inside the pores which in turn catalyzed the growth of carbon nanotubes inside the PECVD system at temperature of 700-750C using mixture of ammonia and acetylene gases. The resulting well-aligned multi-walled carbon nanotubes were further investigated using SEM, TEM and Raman spectroscopy. The size, shape and structure of the grown carbon nanotubes were also discussed.

Performance-Driven Microfabrication-Oriented Methodology for MEMS Conceptual Design with Application in Microfluidic Device Design

Performance and manufacturability are two important issues that must be taken into account during MEMS design. Existing MEMS design models or systems follow a process-driven design paradigm, that is, design starts from the specification of process sequence or the customization of foundry-ready process template. There has been essentially no methodology or model that supports generic, high-level design synthesis for MEMS conceptual design. As a result, there lacks a basis for specifying the initial process sequences. To address this problem, this paper proposes a performance-driven, microfabrication-oriented methodology for MEMS conceptual design. A unified behaviour representation method is proposed which incorporates information of both physical interactions and chemical/biological/other reactions. Based on this method, a behavioural process based design synthesis model is proposed, which exploits multidisciplinary phenomena for design solutions, including both the structural components and their configuration for the MEMS device, as well as the necessary substances for the chemical/biological/other reactions. The model supports both forward and backward synthetic search for suitable phenomena. To ensure manufacturability, a strategy of using microfabrication-oriented phenomena as design knowledge is proposed, where the phenomena are developed from existing MEMS devices that have associated MEMS-specific microfabrication processes or foundry-ready process templates. To test the applicability of the proposed methodology, the paper also studies microfluidic device design and uses a micro-pump design for the case study.

Stimuli-responsive Novel Amphiphilic Polymers for Chemical and Biomedical Applications

Amphiphilic polymers are a class of polymers that self-assemble into different types of microstructure, depending on the solvent environment and external stimuli. Self assembly structures can exist in many different forms, such as spherical micelles, rod-like micelles, bi-layers, vesicles, bi-continuous structure etc. Most biological systems are basically comprised of many of these organised structures arranged in an intelligent manner, which impart functions and life to the system. We have adopted the atom transfer radical polymerization (ATRP) technique to synthesize various types of block copolymer systems that self-assemble into different microstructure when subject to an external stimuli, such as pH or temperature. The systems that we have studied are: (1) pH responsive fullerene (C60) containing poly(methacrylic acid) (PMAA-b-C60); (2) pH and temperature responsive fullerene containing poly[2-(dimethylamino)ethyl methacrylate] (C₆₀-b-PDMAEMA); (3) other responsive water-soluble fullerene systems. By varying temperature, pH and salt concentration, different types microstructure can be produced. In the presence of inorganic salts, fractal patterns at nano- to microscopic dimension were observed for negatively charged PMAA-b-C60, while such structure was not observed for positively charged PDMAEMA-b-C60. We demonstrated that negatively charged fullerene containing polymeric systems can serve as excellent nano-templates for the controlled growth of inorganic crystals at the nano- to micrometer length scale and the possible mechanism was proposed. The physical properties and the characteristics of their self-assembly properties will be discussed, and their implications to chemical and biomedical applications will be highlighted.