GRADIENT-DESCENT BASED UNIT-SELECTION OPTIMIZATION ALGORITHM USED FOR CORPUS-BASED TEXT-TO-SPEECH SYNTHESIS

This paper proposes a gradient-descent based unit selection optimization algorithm for the optimization of unit-cost function weights and for improving the overall performance of the unit-selection algorithm, as used in a corpus-based text-to-speech synthesis system. Complex multidimensional and fuzzy-logic based unit-cost functions are used in the presented unit-selection algorithm. The weights used by these unit-cost functions are usually defined by heuristics or by listening tests. This can be very laborious and time consuming, and does not necessarily result in an optimal performance of the unit-selection algorithm because of multidimensional unit-cost function space, within which different database candidates’ features are evaluated. Using heuristics or listening tests is also rather rigid, especially when working with several different databases or voices. It is especially difficult, within this scope, to set up those weights used in unit-cost functions in order to achieve overall optimal performance of the unit-selection algorithm. The proposed unit-selection optimization process consists of several steps. It is fully automatic, flexible, and fast enough to enable the development of a corpus-based text-to-speech (TTS) system that uses many different voices, without any heuristics or listening tests. This optimization process can also be helpful when evaluating the performances of unit-selection cost functions, and the performance of the unit-selection algorithm itself. The obtained results “suggest” those values that the unit-selection cost-function weights should have in order to obtain smoother transitions between selected unit candidates, after the unit-selection process. The obtained results also hint at the performance level that can be achieved with a given set of unit-cost function weights, and suggest what improvements can be gained when using those additional or changed unit-cost functions included within the unit-selection algorithm.     

The Optimal Sizing of HSS Database in IMS

Our paper deals with the architecture and standards of IP the multimedia subsystem from next generation networks point of view. It describes the structure of the home subscriber server database and then derives equations for sizing of its load in the proposed scenarios. We have used two methods for the database sizing—Erlang C formula and Markov chains. Finally, the achieved results and used methods are evaluated and discussed.

     

Low‐cycle fatigue life of thin‐plate auxetic cellular structures made from aluminium alloy 7075‐T651

In the article, the low‐cycle fatigue life durability of thin‐plate auxetic cellular structures is compared with the thin‐plate specimens of standard shape. Both the re‐entrant auxetic cellular structures and the standard specimens were cut from a 2‐mm‐thick strip of aluminium alloy 7075‐T651. First, a fatigue life curve and a cyclic curve were determined for the standard specimens. A special antibuckling device was applied to prevent the bending of the specimens. The same experimental arrangement was then applied to determine the low‐cycle fatigue life of auxetic structures. In the continuation, the most appropriate method was selected to calculate the fatigue life on the basis of the measured fatigue life curves. Abaqus and SIMULIA fe‐safe software were applied for this purpose. The best predictions for the standard specimens were obtained with the Brown‐Miller method. Finally, the selected method was applied to predict the low‐cycle fatigue life of re‐entrant auxetic cellular specimens.     

A ladder to human-comparable intelligence: an empirical metric

ABSTRACT

The field of Artificial General Intelligence needs a simple and understandable measure of progress to aid in passing the difficult Turing Test. The proposed Ladder Metric introduces a sequence of levels that intelligent agents must climb in order to reach the final, Turing indistinguishable, level. Just like the Turing test, the levels can only be defined in terms of general properties and behaviour rather than specific operational requirements. To track the progress and make comparisons one must climb and determine the achieved rung of the proposed ladder.     

Analytical travel time models for multi aisle automated storage and retrieval systems

Multi aisle automated storage and retrieval systems correspond to multiple picking aisles served by a single storage and retrieval machine. These systems are used in practice when the need for pallet places is high and the throughput capacity is relatively low. Therefore, considerable savings of initial investment costs due to single aisle systems can be achieved using such systems. The objective of the presented study is to develop analytical travel time models for multi aisle automated storage and retrieval systems considering the operating characteristics of the storage and retrieval machine. With the proposed analytical travel time models, average travel time of the storage and retrieval machine, from which the performance of multi aisle automated storage and retrieval systems can be evaluated is determined. A simulation model of multi aisle automated storage and retrieval systems was developed to compare the performances of the proposed analytical travel time models. An example of multi aisle automated storage and retrieval systems with five picking aisles was presented for the evaluation purpose. The results show that proposed analytical travel time models perform satisfactory according to the results of simulation analysis and could be very useful tools for designing of multi aisle automated storage and retrieval systems in real-world applications.     

Effect of plasma treatment on the release kinetics of a chemotherapy drug from biodegradable polyester films and polyester urethane films

Investigation was made into the effect of plasma treatment on the release kinetics of the drug Temozolomide (TMZ) from thin, biodegradable polyester films, comprising polylactic acid (PLA) and polyester urethane. The authors utilized two systems to achieve this, the first being diffuse coplanar surface barrier discharge, applying air as the gaseous medium, while the other involved capacitively coupled radio frequency discharge plasma under an argon atmosphere with hexamethyldisiloxane. Results showed that both forms of plasma treatment positively reduced the undesirable burst effect and benefited the release rate of TMZ. The hydrolytic degradability of the materials was slightly enhanced following hydrophilization, whereas the same diminished after hydrophobization had taken place. This was especially true for PLA due to modification of its wettability.     

Citrinin in cereals and feedstuffs coming from Croatia and Bosnia & Herzegovina

The aim of this study was to investigate the occurrence of citrinin in different cereals (n = 341) and feedstuffs (n = 67) coming from farms and feed factories established in Croatia and Bosnia & Herzegovina. The highest mean citrinin concentration (103 ± 102 µg/kg) was observed in cereals sampled in Bosnia & Herzegovina during 2014, although significantly different levels between the two countries were not observed. Across the cereal samples, the maximal citrinin concentration was determined in wheat (429 µg/kg), while across the feedstuffs, the highest concentration was found in pig feed (63 µg/kg). Despite of the increased citrinin levels found in some samples, especially wheat, the obtained values cannot be compared against the maximum limits, since no such levels are stipulated under the applicable legislation. But, given that data on citrinin are very scarce, they can serve as an indicator of cereal and feed contamination in this part of Europe.     

autoBOT: evolving neuro-symbolic representations for explainable low resource text classification

Machine Learning - Learning from texts has been widely adopted throughout industry and science. While state-of-the-art neural language models have shown very promising results for text...     

Laser Cladding of Cold-Work Tool Steel by Pulse Shaping

Repair welding of cold-work tool steels in cold is very risky and almost impossible by conventional processes. The application of pulse shaping in laser cladding with wire to avoid the solidification problems in relevant steel is demonstrated. The results show that sound remelting and/or cladding can be achieved by the right selection of laser parameters and pulse shape, i.e. long pulse duration, moderate pulse peak powers and ramped-down pulse shape. Despite the defects and softening in the cladding due to the formation of retained austenite, the cladding shows better wear resistance at lower loads compared to the heat-treated base material.