Online learning via congregational gradient descent

We propose and analyse a populational version of stepwise gradient descent suitable for a wide range of learning problems. The algorithm is motivated by genetic algorithms which update a population of solutions rather than just a single representative as is typical for gradient descent. This modification of traditional gradient descent (as used, for example, in the backpropogation algorithm) avoids getting trapped in local minima. We use an averaging analysis of the algorithm to relate its behaviour to an associated ordinary differential equation. We derive a result concerning how long one has to wait in order that, with a given high probability, the algorithm is within a certain neighbourhood of the global minimum. We also analyse the effect of different population sizes. An example is presented which corroborates our theory very well.This work was supported by the Australian Research Council.     

Direction relations and two-dimensional range queries: optimisation techniques

This paper defines direction relations (e.g., north, northeast) between two-dimensional objects and shows how they can be efficiently retrieved using B-, KDB- and R- tree-based data structures. Essentially, our work studies optimisation techniques for 2D range queries that arise during the processing of direction relations. We test the efficiency of alternative indexing methods through extensive experimentation and present analytical models that estimate their performance. The analytical estimates are shown to be very close to the actual results and can be used by spatial query optimizers in order to predict the retrieval cost. In addition, we implement modifications of the existing structures that yield better performance for certain queries. We conclude the paper by discussing the most suitable method depending on the type of the range and the properties of the data.     

Automatic localization of craniofacial landmarks for assisted cephalometry

In this paper we propose a system for localization of cephalometric landmarks. The process of localization is carried out in two steps: deriving a smaller expectation window for each landmark using a trained neuro-fuzzy system (NFS) then applying a template-matching algorithm to pin point the exact location of the landmark. Four points are located on each image using edge detection. The four points are used to extract more features such as distances, shifts and rotation angles of the skull. Limited numbers of representative groups that will be used for training are selected based on k-means clustering. The most effective features are selected based on a Fisher discriminant for each feature set. Using fuzzy linguistics if-then rules, membership degree is assigned to each of the selected features and fed to the FNS. The FNS is trained, utilizing gradient descent, to learn the relation between the sizes, rotations and translations of landmarks and their locations. The data for training is obtained manually from one image from each cluster. Images whose features are located closer to the center of their cluster are used for extracting data for the training set. The expected locations on target images can then be predicted using the trained FNS. For each landmark a parametric template space is constructed from a set of templates extracted from several images based on the clarity of the landmark in that image. The template is matched to the search windows to find the exact location of the landmark. Decomposition of landmark shapes is used to desensitize the algorithm to size differences. The system is trained to locate 20 landmarks on a database of 565 images. Preliminary results show a recognition rate of more than 90%.     

Prediction of residual stress distribution in multi-stacked thin film by curvature measurement and iterative FEA

In this study, residual stress distribution in multi-stacked film by MEMS (Micro-Electro Mechanical System) process is predicted using Finite Element method FEMi. We develop a finite element program for residual stress analysis (RESA) in multi-stacked film. The RESA predicts the distribution of residual stress field in multi-stacked film. Curvatures of multistacked film and single layers which consist of the multi-stacked film are used as the input to the RESA. To measure those curvatures is easier than to measure a distribution of residual stress. To verify the RESA. mean stresses and stress gradients of single and multilayers are measured. The mean stresses are calculated from curvatures of deposited wafer by using Stoney’s equation. The stress gradients are calculated from the vertical deflection at the end of cantilever beam. To measure the mean stress of each layer in multi-stacked film, we measure the curvature of wafer with the left film after etching layer by layer in multi-stacked film.     

Multi-scale qualitative location: A direction-based model

Qualitative locations describe the locations of spatial objects by relating them to a reference frame with qualitative relations. Existing models concerned with regional partitions are mainly topology-based and do not consider the effects of scale changes on locations. This study develops a direction-based multi-scale qualitative location (DMQL) model to fill this gap. First, a cell partition is defined by extending the borders of the minimum bounding rectangles of the regions in a regional partition. Relating spatial objects to all regions by a set of directions is equal to representing the objects as a set of cells in a cell partition. Second, due to the multiple cell representations of spatial objects and the changes in direction relations across scales, some approaches are presented to derive the direction changes between regions in different frames, between spatial objects and regions, and between spatial objects at different scales. Third, the location and relation consistencies of qualitative locations are evaluated based on the cell representations of spatial objects at multiple scales through a case study. The results indicate that the DMQL model can locate objects more precisely than the topology-based models.     

Fabrication and electromagnetic wave absorption property of quartz ceramics with a gradient distribution of BaTiO3

Quartz ceramics with a uniform/gradient distribution of BaTiO₃ (U/G–SO–BTO) are fabricated by cold pressing a powder blend with BTO followed by sintering and using a combined technique of spreading the powder blends with gradually increased BTO layer–by–layer and sintering. The electromagnetic wave absorption properties of these two ceramics are studied in detail. For U–SO–BTO samples, the primary electromagnetic reflection is strong due to the aggravated impedance mismatch at their surfaces. The electromagnetic wave reflectivity of U–SO–BTO could only reach ?7.0?dB when the sample thickness is 6?mm and the BTO content is 8.0?wt%, and it decreases slightly to ?8.1?dB when the sample thickness is increased to 10.0?mm and the BTO content is decreased to 5.0?wt% simultaneously. For G–SO–BTO samples, electromagnetic waves could enter with little reflection due to the weak surface impedance mismatch, and the electromagnetic waves entering these samples could propagate forward while being absorbed gradually with little reflection because of the weak impedance mismatch at the interfaces. The G–SO–BTO samples are promising excellent electromagnetic absorbing materials because their electromagnetic wave reflectivity could reach a level lower than ?12.0?dB and could decrease further from –12.2 to ?13.1?dB as the layer thickness increases from 1.0 to 2.0?mm.     

Gradient microstructure of TC21 alloy induced by hydrogen during hydrogenation

Through melt hydrogenation, a gradient microstructure (α″ + α′)/(α + β_H) has been observed in TC21 alloy. The addition of hydrogen induces martensite transformation and increases the volume fraction of β. It is found that the absorption process of hydrogen atoms can be divided into melting and cooling stages. During cooling, the continuous absorption of hydrogen and the corresponding decrease of freezing point of melt extend solidification time of melt and lead to hydrogen enrichment in the upper of the specimen, which induces the formation of the gradient structure. The hydrogenated TC21 alloy shows higher thermoplasticity compared with the unhydrogenated TC21 alloy. The flow stress of the upper part of the hydrogenated alloy is lower than that of the center part. A gradual variation has been observed in the microhardness along the gradient direction due to variation in the microstructure. The microhardness of the upper surface drops about 45% with 14.6 at.%H.     

A multi-object image segmentation C–V model based on region division and gradient guide

The Chan–Vese (C–V) model is an ineffective method for processing images in which the intensity is inhomogeneous. This is especially true for multi-object segmentation, in which the target may be missed or excessively segmented. In addition, for images with rich texture information, the processing speed of the C–V is slow. To overcome these problems, this paper proposes an effective multi-object C–V segmentation model based on region division and gradient guide. First, a rapid initial contour search is conducted using Otsu’s method. This contour line becomes the initial contour for our multi-object segmentation C–V model based on a gradient guide. To achieve the multi-object segmentation the image is then converted to a single level set whose evolution is controlled using an adaptive gradient. The feasibility of the proposed model is analyzed theoretically, and a number of simulation experiments are conducted to validate its effectiveness.     

Effect of transition layer on the performance of hydroxyapatite/titanium nitride coating developed on Ti-6Al-4V alloy by magnetron sputtering

A gradient transition multilayer hydroxyapatite/titanium nitride (HA/TiN) coating was prepared on the Ti-6Al-4V alloy by magnetron sputtering. The composition, surface topography, microstructure, adhesion strength and electrochemical properties of the as-deposited coatings were characterized by SEM/EDS, AFM, XRD, FT-IR and electrochemical workstation. The experimental results showed that the single TiN coating deposited at a partial pressure of nitrogen (N₂) of 0.08?Pa had the best internal stress and tribological performance, and its volume loss was only 0.89% of that of Ti-6Al-4V alloy. The introduction of the TiN transition layer greatly improved the wear resistance of the Ti-6Al-4V alloy, and the adhesion strength of the HA layer to the substrate increased from 6.50?±?0.5?N to 11.70?±?1.2?N, an increase of 56%. The HA/TiN coating surface consisted of uniform hemispherical particles with dense structure and invisible defects (micro-cracks and pores). For the HA surface layer, the crystal structure and active hydroxyl (-OH) group was restored after heat treatment. Potentiodynamic polarization experiments indicated that the HA/TiN coating achieved the lowest corrosion current density and the most positive corrosion potential compared to the single TiN layer and Ti-6Al-4V alloy. In summary, it can be conclude that the gradient transition layer can well improve the mechanical properties and electrochemical behavior of the titanium alloy, and largely ensuring the stability of the surface bioactive coating.     

Influence of a hard transition layer on the microstructure and properties of diamond-like carbon/hydroxyapatite composite coating prepared by magnetron sputtering

The nanostructured diamond-like carbon/hydroxyapatite composite coating (DLC/HA) was deposited using magnetron sputtering technique with a densely packed columnar cross-sectional structure and a uniform granular surface morphology. After heat treatment, the amorphous structure of the coating was transformed into a crystal structure. Nanohardness and scratch tests results demonstrated the DLC transition layer significantly enhanced the nanohardness of Ti6Al4V substrates from 4.8 GPa to 10.4 GPa, and increased critical load from 16.6 N (pure HA layer) to 26.5 N (DLC layer) without obvious brittle fracture, flaking and delamination. Electrochemical and immersion tests results demonstrated that DLC/HA composite coatings with a dense gradient transition interlayer had better corrosion resistance and could prevent harmful metal ions being released into the SBF solution more effectively than single HA coatings. Furthermore, active Ca²⁺ ions can be rapidly released from the coating surface during initial immersion in the SBF solution, and facilitated the formation of bone-like apatite.