Acoustic features extraction for emotion recognition

In the last decade, the efforts of spoken language processing have achieved significant advances, however, the work with emotional recognition has not progressed so far, and can only achieve 50% to 60% in accuracy. This is because a majority of researchers in this field have focused on the synthesis of emotional speech rather than focusing on automating human emotion recognition. Many research groups have focused on how to improve the performance of the classifier they used for emotion recognition, and few work has been done on data pre-processing, such as the extraction and selection of a set of specifying acoustic features instead of using all the possible ones they had in hand. To work with well-selected acoustic features does not mean to delay the whole job, but this will save much time and resources by removing the irrelative information and reducing the high-dimension data calculation. In this paper, we developed an automatic feature selector based on a RF2TREE algorithm and the traditional C4.5 algorithm. RF2TREE applied here helped us to solve the problems that did not have enough data examples. The ensemble learning technique was applied to enlarge the original data set by building a bagged random forest to generate many virtual examples, and then the new data set was used to train a single decision tree, which selects the most efficient features to represent the speech signals for the emotion recognition. Finally, the output of the selector was a set of specifying acoustic features, produced by RF2TREE and a single decision tree.

OWDEAH: Online Web Data Extraction based on Access History

Web data extraction systems are the kernel of information mediators between users and heterogeneous Web data resources. How to extract structured data from semi-structured documents has been a problem of active research. Supervised and unsupervised methods have been devised to learn extraction rules from training sets. However, trying to prepare training sets (especially to annotate them for supervised methods), is very time-consuming. We propose a framework for Web data extraction, which logged usersrsquo access history and exploit them to assist automatic training set generation. We cluster accessed Web documents according to their structural details; define criteria to measure the importance of sub-structures; and then generate extraction rules. We also propose a method to adjust the rules according to historical data. Our experiments confirm the viability of our proposal.

Extraction of compounds associated with water repellency in sandy soils of different origin

After an initial evaluation of several solvents, the efficiency of Soxhlet extractions with isopropanol/ammonia (s.g. 0.88) (70 : 30 v : v; 24 h) in extracting compounds associated with water repellency in sandy soils was examined using a range of repellent and wettable control soils (n = 15 and 4) from Australia, Greece, Portugal, The Netherlands, and the UK. Extraction efficiency and the role of the extracts in causing soil water repellency was examined by determining extract mass, sample organic carbon content and water repellency (after drying at 20°C and 105°C) pre- and post-extraction, and amounts of aliphatic C–H removed using DRIFT, and by assessing the ability of extracts to cause repellency in acid-washed sand (AWS).

Key findings are: (i) none of organic carbon content, amount of aliphatic C–H, or amount of material extracted give any significant correlation with repellency for this diverse range of soils; (ii) sample drying at 105°C is not necessarily useful before extraction, but may provide additional information on extraction effectiveness when used after extraction; (iii) the extraction removed repellency completely from 13 of the 15 repellent samples; (iv) extracts from all repellent and wettable control soils were capable of inducing repellency in AWS. The findings suggest that compounds responsible for repellency represent only a fraction of the extract composition and that their presence does not necessarily always cause repellency.

The application of rough set and Kohonen network to feature selection for object extraction

Selecting a set of features which is optimal for a given task is a problem which plays an important role in a wide variety of contexts including pattern recognition, images understanding and machine learning. The paper describes an application of rough sets method to feature selection and reduction in texture images recognition. The proposed methods include continuous data discretization based on Kohonen neural network and maximum covariance, and rough set algorithms for feature selection and reduction. The experiments on trees extraction from aerial images show that the methods presented in this paper are practical and effective.

Automatic pattern-taxonomy extraction for web mining

In this paper, we propose a model for discovering frequent sequential patterns, phrases, which can be used as profile descriptors of documents. It is indubitable that we can obtain numerous phrases using data mining algorithms. However, it is difficult to use these phrases effectively for answering what users want. Therefore, we present a pattern taxonomy extraction model which performs the task of extracting descriptive frequent sequential patterns by pruning the meaningless ones. The model then is extended and tested by applying it to the information filtering system. The results of the experiment show that pattern-based methods outperform the keyword-based methods. The results also indicate that removal of meaningless patterns not only reduces the cost of computation but also improves the effectiveness of the system.

Extraction and classification of self-consumable sport video highlights using generic HMM

This paper aims to automatically extract and classify self-consumable sport video highlights. For this purpose, we will emphasize the benefits of using play-break sequences as the effective inputs for HMMbased classifier. HMM is used to model the stochastic pattern of high-level states during specific sport highlights which correspond to the sequence of generic audio-visual measurements extracted from raw video data. This paper uses soccer as the domain study, focusing on the extraction and classification of goal, shot and foul highlights. The experiment work which uses183 play-break sequences from 6 soccer matches will be presented to demonstrate the performance of our proposed scheme.

Object extraction using genetic algorithms and morphological processing

This paper introduces a novel methodology for object detection using genetic algorithms and morphological processing. The method employs a kind of object oriented structuring element, which are derived using genetic algorithm operating. The population of morphological filters iteratively evaluated according to a statistical performance index corresponding to object extraction ability, and evolves into an optimal structuring elements using the evolution principles of genetic search. Experimental results of object extraction in high resolution satellite images are presented to illustrate the merit and feasibility of the proposed method.

Is water extraction a threat to insect populations in intermittently-flowing streams?

Stream insects use a variety of refuges to survive drought that vary in their vulnerability to water extraction. We surveyed drought refuges in 16 ephemeral streams with different flow regimes. Perennial sections were the most important drought refuge, but are the most likely to disappear when water is extracted.

Competition and unitization in oil extraction : a tale of two tragedies

This paper compares two organizational modes for extracting oil from a common pool: unitization and competitive extraction. The analysis suggests that the general presumption that unitization is surplus enhancing relative to competitive extraction may not always be valid due to contractual incompleteness associated with unitization contracts. While competitive extraction suffers from the tragedy of the commons, unitization can be subject to the dual tragedy of the anticommons. This provides an explanation for the puzzle confronting the oil industry that firms are often reluctant to voluntarily enter unitization agreements.

Critical assessment of various techniques for the extraction of carotenoids and co-enzyme Q10 from the Thraustochytrid strain ONC-T18

A variety of techniques for extracting carotenoids from the marine Thraustochytrium sp. ONC-T18 was compared. Specifically, the organic solvents acetone, ethyl acetate, and petroleum ether were tested, along with direct and indirect ultrasonic assisted extraction (probe vs bath) methods. Techniques that used petroleum ether/acetone/water (15:75:10, v/v/v) with 3 h of agitation, or 5 min in an ultrasonic bath, produced the highest extraction yields of total carotenoids (29−30.5 μg g^-1). Concentrations up to 11.5 μg g^-1 of canthaxanthin and 17.5 μg g^-1 of β-carotene were detected in extracts stored for 6 weeks. Astaxanthin and echinenone were also detected as minor compounds. Extracts with and without antioxidants showed similar carotenoid concentration profiles. However, total carotenoid concentrations were approximately 8% higher when antioxidants were used. Finally, an easy-to-perform and inexpensive method to detect co-enzymes in ONC-T18 was also developed using silica gel TLC plates. Five percent methanol in toluene as a mobile phase consistently eluted co-enzyme Q10 standards and could separate the co-enzyme fractions present in ONC-T18.

Evaluation of fatty acid extraction methods for Thraustochytrium sp. ONC-T18

Various extraction methods were assessed in their capacity to extract fatty acids from a dried biomass of Thraustochytrium sp. ONC-T18. Direct saponification using KOH in ethanol or in hexane:ethanol was one of the most efficient techniques to extract lipids (697 mg g^-1). The highest amount of fatty acids (714 mg g^-1) was extracted using a miniaturized Bligh and Dyer extraction technique. The use of ultrasonics to break down cell walls while extracting with solvents (methanol:chloroform) also offered high extraction yields of fatty acids (609 mg g^-1). Moreover, when the transesterification mixture used for a direct transesterification method was doubled, the extraction of fatty acids increased approximately 77% (from 392 to 696 mg g^-1). This work showed that Thraustochytrium sp. ONC-T18 has the ability to produce over 700 mg g^-1 of lipids, including more than 165 mg g-1 of docosahexaenoic acid, which makes this microorganism a potential candidate for the commercial production of polyunsaturated fatty acids. Finally, other lipids, such as myristic, palmitic, palmitoleic, and oleic acids, were also produced and recovered in significant amounts (54, 196, 123, and 81 mg g^-1), respectively.

Characterisation of nickel(II) extraction by 2-hydroxy-5-nonylacetophenone oxime (LIX 84) in a micellar phase

The properties of the nickel(II)/2-hydroxy-5-nonylacetophenone oxime (HNAPO), an active ingredient in LIX 84, extraction system were characterised in a micellar system. The extinction coefficient, λ_max of HNAPO (316 nm) and the Ni²⁺ complex (387 nm) in a neutral micellar system, poly dispersed octa-ethyleneglycol mono-n-dodecyl ether (G₁₂A₈) were determined as 3100 and 3500 M⁻¹ cm⁻¹, respectively. HNAPO was found to have a neutral micellar phase and bulk aqueous phase pK_a of 11.5 and 12.5, respectively. The extraction equilibrium constant, K_ex, was determined to be 10^−8.0, and the deviation from theory observed at high pH can be accounted for by consideration of the competition for nickel(II) ions by hydroxide ions and HNAPO. A micellar phase of octa-ethyleneglycol mono-n-dodecyl ether (C₁₂E₈) was determined to be an appropriate model of the free oil/water interface from the solubilised location of HNAPO. Utilising the interfacial probe, 4-heptadecyl-7-hydroxy coumarin (HHC) allowed the determination of the electrostatic surface potential of mixed micelles of G₁₂A₈ and sodium dodecyl sulphate (SDS) or dodecyl trimethyl ammonium chloride (DTAC). The electrostatic surface potential was a linear function of the number of additional surfactant monomers within the G₁₂A₈ micelle, for the concentration range studied. For G₁₂A₈/DTAC mixed micelles, the surface potential was given by +1.1 mV per DTAC molecule per micelle, and for G₁₂A₈/SDS mixed micelles the relationship was −1.4 mV per SDS molecule per micelle.

Improved 3D thinning algorithms for skeleton extraction

In this study, we focused on developing a novel 3D Thinning algorithm to extract one-voxel wide skeleton from various 3D objects aiming at preserving the topological information. The 3D Thinning algorithm was testified on computer-generated and real 3D reconstructed image sets acquired from TEMT and compared with other existing 3D Thinning algorithms. It is found that the algorithm has conserved medial axes and simultaneously topologies very well, demonstrating many advantages over the existing technologies. They are versatile, rigorous, efficient and rotation invariant.