On the statistical physics of radial basis function networks

Techniques from statistical physics have been applied successfully in recent years to the analysis of the generalization performance of neural networks. However, most of the analysis to date has been for perceptron-like networks or simple generalizations thereof such as committee machines, and none of the networks studied are used to any significant extent in practice. This letter presents results obtained in applying techniques from statistical physics to a popular class of neural networks that has been used successfully in many practical applications: the Gaussian radial basis function networks. We obtain expressions for the learning curves exhibited by these networks in the high-temperature limit for both realizable and unrealizable target rules.     

Structure‐independent motion recovery from a monocular image sequence with low fill fraction

Real‐world image sequences are usually characterized by frequent loss and replacement of tracked feature data, and the key to accurate motion recovery from such image sequences lies in the synthesis of a structure‐independent dynamic filter that enables the replacement of lost point features with newly acquired points in a seamless manner. In this note, a novel nonlinear observer is proposed that relies on filtered estimates of optical flow to accomplish structure‐independent motion recovery from monocular image sequences with a low fill fraction. With a single component of linear velocity assumed known, the proposed scheme relies on the perspective observation of at least five points to yield exponentially convergent estimates of the unknown motion parameters that converge to a uniform, ultimate bound in the presence of model error. The unknown linear and angular velocities are assumed to be generated using an imperfectly known model that incorporates a bounded uncertainty, and optical flow estimation is accomplished using a robust differentiator that is based on the sliding‐mode technique. Numerical results are used to validate and demonstrate superior observer performance compared to a leading alternative design on a real‐world–like image sequence that is characterized by significant measurement noise and high feature turnover rate.     

An Optical Flow-Based Solution to the Problem of Range Identification in Perspective Vision Systems

A classical problem in machine vision is the range identification of an object moving in three-dimensional space from the two-dimensional image sequence obtained with a monocular camera. This study presents a novel reduced-order optical flow-based nonlinear observer that renders the proposed scheme suitable for depth estimation applications in both well-structured and unstructured environments. In this study, a globally exponentially stable observer is synthesized, where optical flow estimates are derived from tracking feature trajectory on the image plane over successive camera frames, to yield asymptotic estimates of feature depth at a desired convergence rate. Furthermore, the observer is shown to be finite-gain \(\mathcal {L}_{p}\) stable ?p∈[1,∞] in the presence of exogenous disturbance influencing camera motion, and is applicable to a wider class of perspective systems than those considered by alternative designs. The observer requires minor apriori system information for convergence, and the convergence condition arises in a natural manner with an apparently intuitive interpretation. Numerical and experimental studies are used to validate and demonstrate robust observer performance in the presence of significant measurement noise.     

A data-based framework for fault detection and diagnostics of non-linear systems with partial state measurement

A novel framework based on the use of dynamic neural networks for data-based process monitoring, fault detection and diagnostics of non-linear systems with partial state measurement is presented in this paper. The proposed framework considers the presence of three kinds of states in a generic system model: states that can easily be measured in real time and in-situ, states that are difficult to measure online but can be measured offline to generate training data, and states that cannot be measured at all. The motivation for such a categorization of state variables comes from a wide class of problems in the manufacturing and chemical industries, wherein certain states are not measurable without expensive equipments or offline analysis while some other states may not be accessible at all. The framework makes use of a recurrent neural network for modeling the hidden dynamics of the system from available measurements and uses this model along with a non-linear observer to augment the information provided by the measured variables. The performance of the proposed method is verified on a synthetic problem as well as a benchmark simulation problem.     

Hyperbranched polyurethane/Fe3O4 thermosetting nanocomposites as shape memory materials

Nanocomposites of hyperbranched polyurethane were prepared by the in situ pre-polymerization technique with Fe₃O₄ nanoparticles. The synthesized Fe₃O₄ nanoparticles were characterized by the Fourier transform infrared spectroscopy and the X-ray diffraction study. The transmission electron microscopic study indicates the homogeneous distribution of Fe₃O₄ nanoparticles in the polymer matrix. The mechanical, thermal and shape memory behaviors of the nanocomposites were studied as a function of nanomaterial content. The glycidyl bisphenol-A based epoxy cured thermosetting nanocomposites exhibited significant improvement of tensile strength (5.7–18 MPa), scratch hardness (3.0–6.5 kg) and thermal stability (241–275 °C) with the increase of the content of Fe₃O₄. The nanocomposites possess excellent shape fixity over the repeated cycles of test. They also showed good shape recovery under the application of microwave energy. The shape recovery speed found to increase with the increase of the loading of Fe₃O₄ in the nanocomposites. Thus, the prepared nanocomposites might be utilized as advanced shape memory materials in their potential fields.     

Mesua ferrea L. seed oil based highly branched polyester/epoxy blends and their nanocomposites

Mesua ferrea L. seed oil based highly branched polyester and epoxy resins blends were prepared by mechanical mixing at different weight ratios. The best performing blend was used as the matrix for the preparation of nanocomposites with different dose levels of organophilic montmorillonite (OMMT) nanoclay. The prepared nanocomposites were characterized by X‐ray diffraction, scanning electron microscopy, Fourier transform infrared spectroscopy, and transmission electron microscopy. Data resulting from the mechanical and thermal studies of the blends and nanocomposites indicated improvements in the tensile strength and thermal stability to appreciable extents for the nanocomposites with OMMT loading. The nanocomposites were characterized as well‐dispersed, partially exfoliated structures with good interfacial interactions. From the X‐ray diffraction analysis, the absence of d₀₀₁ reflections of the OMMT clay in the cured nanocomposites indicated the development of an exfoliated clay structure, which was confirmed by transmission electron microscopy. The homogeneous morphologies of the pure polyester/epoxy blend and clay hybrid systems were ascertained with scanning electron microscopy. The tensile strength of the 5 wt % clay‐filled blend nanocomposite system was increased by 2.4 times compared to that of the pure blend resin system. The results suggest that the prepared nanocomposites have the potential to be used as active thin films for different applications. © 2011 Wiley Periodicals, Inc. J Appl Polym Sci, 2011     

High-Temperature Indentation Studies on the {110} Plane of Single-crystal Magnesium Oxide

The indentation behavior (Vickers) of Single-crystal MgO was studied as a function of temperature (20° to 1000°C). Indentations were made on the {110} plane, with the indents oriented such that one indent diagonal was parallel to the 〈001〉 direction. Using etchant techniques, the dislocation etch pit structures were examined both in the plane of the indentation and in cross section. All the observed slip traces were found to be consistent with primary slip ({110}〈 1 10〉), with no evidence of secondary slip, even at 1000°C. Radial cracking was observed only at the pair of indent corners joined by the indent diagonal parallel to 〈001〉. The crack length increased with temperature ( T ) for indentations conducted at T < 800°C. For indents made at 800°C or higher, however, no cracking occurred. These results are discussed both with respect to an existing slip-induced crack nucleation model, and the change in crack driving force and toughness with indentation temperature.     

Enzyme‐assisted water‐extraction of oil and protein from rice bran

Enzymatic water‐extraction of oil and proteins from rice bran was studied in a laboratory‐scale set‐up. The effects of the following enzymes – Celluclast 1.5L, hemicellulase, Pectinex Ultra SP‐L, Viscozyme L, Alcalase 0.6L and papain – on oil and protein extraction yields, and the level of reducing sugars in the extract were investigated. The results showed that Alcalase was most effective in enhancing oil and protein extraction yields. Papain was found to be superior to all carbohydrase enzymes but it gave lower yields than Alcalase. Celluclast 1.5L, hemicellulase, Pectinex Ultra SP‐L and Viscozyme L did not affect yields significantly but increased the level of reducing sugars in the extract. © 2002 Society of Chemical Industry     

Bio-based elastomeric hyperbranched polyurethanes for shape memory application

The bio-based shape memory hyperbranched polyurethanes (HBPUs) have attracted tremendous attention both from academic and industrial researchers due to their strong potential in biomedical and other advanced applications. In the present investigation HBPUs have been synthesized from poly(??-caprolactone)diol as a macroglycol, butanediol as a chain extender, triethanolamine as a branch generating moiety, monoglyceride of Mesua ferrea L. seed oil as a bio-based chain extender, at different percentages and toluene diisocyanate by a two step one pot A₂?+?B₃ approach. The structure of the synthesized hyperbranched polyurethane was characterized by FTIR, ^IH NMR, XRD and SEM studies. ¹H NMR study indicates the formation of highly branched structure with degree of branching 0.93 for polyurethane with 5?wt% monoglyceride. TGA results indicated the increment of thermal stability from 185 to 240?°C with the increase of monoglyceride content from (0?C15) wt% for the HBPUs. The shape memory effect of the hyperbranched polyurethane increased with the increase of monoglyceride in the polymer. However, mechanical properties like tensile strength and elongation at break decreased from 19.31 to 11.48?MPa and 835 to 497%, respectively, with the increase in amount of bio-based component. Excellent impact strength and very good chemical resistance were also observed for the hyperbranched polymers. The studied bio-based HBPUs exhibit excellent shape fixity (95?C99)% as well as shape recovery 100%. Thus, the studied HBPUs have the potential to be used as advanced shape memory materials.     

Renewable resource modified polyol derived aliphatic hyperbranched polyurethane as a biodegradable and UV‐resistant smart material

Renewable resource tailored tough, elastomeric, biodegradable, smart aliphatic hyperbranched polyurethanes were synthesized using castor oil modified polyol containing fatty amide triol, glycerol, diethanolamine and monoglyceride of sunflower oil via an A_x + B_y (x , y ≥ 2) approach. To the best of our knowledge, this is the first report of the synthesis of solely aliphatic hyperbranched polyurethanes by employing renewable resources. The synthesized polyurethanes were characterized by Fourier transform infrared, NMR and XRD techniques. The hyperbranched polyurethanes exhibited good mechanical properties, especially elongation at break (668%), toughness (32.16 MJ m^?3) and impact resistance (19.02 kJ m^?1); also high thermal stability (above 300 °C) and good chemical resistance. Also, the hyperbranched polyurethanes were found to show adequate biodegradability and significant UV light resistance. Moreover, they demonstrated excellent multi‐stimuli‐driven shape recovery ability (up to 97%) under direct sunlight (10⁵ lux), thermal energy (50 °C) and microwave irradiation (450 W). The performance of the hyperbranched polyurethanes was compared with renewable resource based and synthetic linear polyurethane to judge the superiority of the hyperbranched architecture. Therefore, these new aliphatic macromolecules hold significant promise as smart materials for advanced applications. © 2017 Society of Chemical Industry