DJ el Niño: expressing synthetic emotions with music

The purpose of this work is twofold: (1) to present an artistic experiment on how to use artificial intelligence to develop a different kind of DJ, and (2) to test a cognitive model on how music expresses emotions. Based on a former model conceived by the author, electronic music loops were tagged according to the type and intensity of the expressed emotion. Then, using a feedback model, an artificial personality was arranged, which was affected by the music and played the loops depending on the emotional state the artificial personality was in. The efficiency and credibility of the model was tested in a live event during the Z2000 art festival in Berlin, with encouraging results.

---

A CAVE-like environment as a tool for full-size train design

Simulations of models, in all different areas, is an expanding, attractive line of work. More and more applications are taking advantage of the improvements in technology and knowledge in this field, thus achieving results that would have been impossible to achieve with a real model, or foreseeing facts that otherwise would have been encountered too late in the production process. The rail industry is one possible beneficiary of this approach. Usually, before commencing the fabrication process of a new train, the construction of a full-size model is mandatory. Instead of building this full-size real model, which leaves little room for later, last-minute modifications, a virtual model can be built in the digital realm, thus offering a new platform for easier interaction with it. In this article, a simulation of a train is presented in order to tackle visual, aesthetic and ergonomic issues. The simulation runs on a PC-based CAVE-like architecture, offering a certain degree of interaction to the user, and combines static and dynamic computer-generated imagery, both with and without stereoscopy for 3D visualisation, as well as augmented virtuality techniques for the integration of the train with its environment.     

New learning rules for the ASSOM network

The ASSOM is a self-organising neural network with the capability of adapting to linear subspaces. Here we propose two new methods to train the ASSOM network. A nonlinear system of equations is derived for network training. This system can be solved by a gradient-based approach or by the Levenberg–Marquardt method. Each of these two approaches gives a different learning rule. A comparison is carried out among the original Kohonens method and the proposed learning rules. Experimental results are reported, including a convergence speed experiment and a speech processing application, which show that the new learning rules have better performance than the original one.     

A regression model describing the effect of pH,NaCl and temperature on<Emphasis Type="Italic"> D</Emphasis> values of<Emphasis Type="Italic"> Bacillus stearothermophilus</Emphasis> spores

A new microbiological predictive model was developed, relating pH, NaCl concentration and temperature (T) with the decimal reduction coefficient D as dependent variable, using data previously reported by [1]. The model was obtained by multiple linear regression using only three predictor variables (T ², pH² and NaCl), easily computed from the basic variables pH, NaCl concentration and temperature (T). The fitted model is robust, stable and satisfies all the basic hypotheses of the regression models. It also provides simplicity when compared with previously published models based on multiple linear regression tools.     

Simulation of multistage extraction of antioxidants from Chilean hazelnut (<Emphasis Type="Italic">Gevuina avellana</Emphasis>) hulls

The objective of this study was to analyze the factors affecting the ethanol extraction of Chilean hazelnut (Gevuina avellana) hulls to obtain antioxidant compounds. The effects of temperature on the kinetics of polyphenolics extraction and on the antioxidant activities of the extracts were assessed. The radical-scavenging activities of the extracts were comparable with that of BHA when used at the same concentration. The optimal temperature for the antioxidant activities of the extracts was 40°C. A four-stage cross-flow extraction was carried out and a four-stage countercurrent extraction was simulated, where each stage lasted 30 min. Best results were obtained with countercurrent extraction, which produced an extract that showed 94.4% α,α-diphenyl-β-picrylhydazyl radical inhibition, compared with 86.2% obtained in the first stage of cross-flow extraction and 92.62% in a batch extraction that lasted 100 h. UV-vis and NIR spectra of extracts from cross-flow and from the simulated countercurrent extraction revealed that the composition of extracts varied along the stages and was affected by the operational strategy.     

Solid-phase microextraction liquid chromatography/tandem mass spectrometry to determine postharvest fungicides in fruits

A method to determine five postharvest fungicides (dichloran, flutriafol, o-phenylphenol, prochloraz, tolclofos methyl) in fruits (cherries, lemons, oranges, peaches) has been developed using solid-phase microextraction (SPME) coupled to liquid chromatography (LC) with photodiode array (DAD), mass spectrometry (MS), or tandem mass spectrometry (MS/MS) with ion trap detection. Extraction involved sample homogenization with an acetone/water solution (5:1), filtration, and acetone evaporation prior to fiber extraction. The pesticides were isolated with a fused-silica fiber coated with 50-microm Carbowax/template resin. The effects of pH, ion strength, sample volume, and extraction time were investigated, and their impact on the SPME-LC/MS was studied. Dynamic and static modes of desorption were compared and the variables affecting desorption processes in SPME-LC optimized. Static desorption provided the best recoveries and peak shapes. Recoveries at the limit of quantification (LOQ) levels were between 10% for prochloraz and 60% for o-phenylphenol, with relative standard deviations from 13.6% for prochloraz to 3.1% for o-phenylphenol. The versatility of the method was also exhibited by its excellent linearity in the concentration intervals between 0.0005 and 5 mg kg(-1) for dichloran and 0.01-10 mg kg(-1) for tolclofos methyl and prochloraz. LOQs ranged from 0.25 to 1 microg g(-1) using DAD, from 0.002 to 0.01 microg g(-1) using LC/MS, and from 0.0005 to 0.01 to microg g(-1) using LC/MS/MS. LOQs obtained in the present study using LC/MS and LC/MS/MS are lower than maximum residue limits established for all the fungicides in any matrix studied. The method enables to determine polar pesticides at low-microgram per gram levels in fruits.     

A contribution to understanding secondary reactions in linear alkylbenzene sulfonation

Modern industrial sulfonation of linear alkylbenzene (LAB) in falling film reactors produces p-alkylbenzene sulfonic acid (HLAS), which is usually neutralized with caustic soda (NaOH) and used in detergent formulations. During the sulfonation process, other products such as anhydrides and sulfones are also formed. Four reactions are proposed to occur during aging and hydrolysis. As a consequences, approximately 25% of sulfones and 75% of unreacted LAB are removed during aging and are transformed into additional active matter. Anhydrides are completely eliminated during hydrolysis. On the other hand, sulfones formed during the process are difficult to remove once formed. The goal of this work was to optimize the various operating conditions of the sulfonation process and to understand the various secondary reactions that occur therein so as to obtain a maximal active ingredient concentration in the final sulfonated product and a minimum of unsulfonated matter. Formation of sulfones can be minimized by using an SO₃/LAB molar ratio slightly lower than the theoretical optimum.     

An Object-oriented Interpretation of the EAT System

In a previous paper we characterized, in the Category Theory setting, a class of implementations of Abstract Data Types, which has been suggested by the way of programming in the EAT system. (EAT, Effective Algebraic Topology, is one of Sergeraerts systems for effective homology and homotopy computation.) This characterization was established using classical tools, in an unrelated way to the current mainstream topics in the field of Algebraic Specifications. Looking for a connection with these topics, we have found, rather unexpectedly, that our approach is related to some object-oriented formalisms, namely hidden specifications and the coalgebraic view. In this paper, we explore these relations making explicit the implicit object-oriented features of the EAT system and generalizing the data structure analysis we had previously done.     

Differentiating random upper semicontinuous functions under the integral sign

In this paper we analyze sufficient conditions to guarantee the differentiability under the integral sign of a certain class of random upper semicontinuous functions depending on a real-valued parameter. Several concepts of differential for this mappings are considered: the Fréchet differentiability of the support function, the π-differentiability, the De Blasi differentiability, and thes-differentiability.     

On discrete multivariate distributions symmetric in frequencies

In this paper we describe a new class of discrete multivariate distributions which verify that their probability mass function is invariant when their univariate variables are permuted. These distributions may be generated by a multivariate extension of the Gauss function₂ F ₁ with matrix argument. A methodology that permits the fit of these distributions to real data is developed. A fit of a distribution for bivariate real data is shown and is compared with fits obtained by means of other usual bivariate distributions generated by extensions of the Gauss function.