A privacy-preserving distributed transfer learning in activity recognition

Telecommunication Systems - Along with the widespread use of smartphones, activity recognition using embedded inertial sensors has intrigued researchers. The learning and employing activity...     

An evolutionary decoding method for HMM-based continuous speech recognition systems using particle swarm optimization

The main recognition procedure in modern HMM-based continuous speech recognition systems is Viterbi algorithm. Viterbi algorithm finds out the best acoustic sequence according to input speech in the search space using dynamic programming. In this paper, dynamic programming is replaced by a search method which is based on particle swarm optimization. The major idea is focused on generating initial population of particles as the speech segmentation vectors. The particles try to achieve the best segmentation by an updating method during iterations. In this paper, a new method of particles representation and recognition process is introduced which is consistent with the nature of continuous speech recognition. The idea was tested on bi-phone recognition and continuous speech recognition workbenches and the results show that the proposed search method reaches the performance of the Viterbi segmentation algorithm ; however, there is a slight degradation in the accuracy rate.     

Development of core to solve the multidimensional multiple-choice knapsack problem

The multidimensional multiple-choice knapsack problem (MMKP) is an extension of the 0–1 knapsack problem. The core concept has been used to design efficient algorithms for the knapsack problem but the core has not been developed for the MMKP so far. In this paper, we develop an approximate core for the MMKP and utilize it to solve the problem exactly.     

An extended KCF tracking algorithm based on TLD structure in low frame rate videos

The KCF (Kernelized Correlation Filter) algorithm achieved a good performance on target tracking challenges. However, it still has some defects and problems of false tracking in low frame rate (LFR) scenarios, target scale variation, occlusion and out of view target, that exists in the correlation filter based methods. In this paper, we overcome the shortcomings of KCF tracking algorithm based on Tracking-Learning-Detection (TLD) framework. The proposed algorithm trained two classifiers simultaneously, based on semi supervised co-training learning algorithm. Then, we comparatively evaluate the proposed method on TB-100 datasets by other trackers. The experimental results demonstrate that the precision and robustness of the improved tracking algorithm is higher than traditional KCF, TLD and the other top state-of-the-art tracking algorithms in LFR videos.

     

Decision fusion of horizontal and vertical trajectories for recognition of online Farsi subwords

Online handwriting is formed by a combination of horizontal and vertical trajectories. If these trajectories are treated separately, new recognition methods are emerged. In contrast, one classifier is often used to recognize handwriting. In this work, some features for x(t) and y(t) signals were proposed and used to make two separate classifiers. After initial recognition by these classifiers, their results were fused for final recognition. Using HMM classifiers and simple product rule for decision fusion, the recognition results of 42 classes of Farsi subwords showed promising achievements.     

GVSEE: a new global model to estimate software services development effort

As software services have become a main part of companies in recent years, accurate and efficient estimates of required effort for their development has turned into a major concern. Furthermore, the high diversity, complexity, non-normality, and inconsistency of software services have made an estimation of the needed development effort a very difficult task. In spite of the numerous studies conducted, and improvements made, in the past, no single model has yet been introduced that can reliably estimate the required effort. Therefore, apparently, it is impossible to introduce a global and efficient model for all types of services. This article proposes a new model called Global Village Service Effort Estimator that emphasizes the idea ‘Think locally, act globally.’ Unlike previous studies, this model does not rely on a specific method and, in addition to combining methods, takes a local look at each software service with the help of fuzzy clustering. The model was evaluated on a real data-set International Software Benchmarking Standard Group and on two artificial data-sets, and obtained results indicated its tangible efficiency and the lack of accuracy of other models. Besides its greater accuracy, other advantages of the proposed model over the other models are its adaptability and flexibility in confronting complexities and uncertainties.     

Determination of water-soluble vitamins in soft drinks and vitamin supplements using capillary electrophoresis

A method for the determination of six water-soluble vitamins based on capillary electrophoresis (CE) operated in micellar mode was developed. Thiamine hydrochloride (vitamin B1), riboflavin (vitamin B2), pyridoxine hydrochloride (vitamin B6), pantothenic acid (vitamin B5), nicotinamide (vitamin B3), and cobalamin (Vitamin B12) could be separated in a single run. All CE parameters such as buffer composition and operation temperature were optimized in order to achieve better separation. Relative standard deviations (RSDs) of the described method ranged from 1.08 to 3.68% (intra-day precision) and 1.26 to 3.35% (inter-day precision). The method was then used for measuring various soft drinks and vitamin supplements directly without any step of sample cleanup. The determination of niacin was successful for all samples tested, reaching recoveries near 100%. Riboflavin and pyridoxine were quantified successfully in some but not all samples. Therefore, an evaluation on a case-by-case basis is mandatory. When applicable, this method provides a fast, accurate, simple, and inexpensive way to quantify selected vitamins, and is therefore well suited for routine analysis in soft drink industry.     

Improving the Performance of HF Radio Networks in the Presence of Interference through Automatic Link Establishment with Frequency Hopping Technique

In the cement industry, a rotary kiln is a pyro-processing device that is used to measure temperature. Measuring and maintaining a certain range of temperature in the rotary kiln is important to ensure the production of quality clinker granules. The assessment of consuming zone temperature is acquired using radiation pyrometers from the temperature of a hotspot. However, it is a difficult task to measure the burning zone temperature due to the very high temperature developed in the turning furnace sintering process. Existing pyrometer and camera based techniques are not able to provide accurate temperature and temperature variations developed in the burning zone. This research work considers flame image processing using region of interest (ROI), fuzzy logic, and neural networks for efficient temperature measurement. Various temperature measurement and control techniques are utilized in the existing conventional (Prasanna and Bojja in ESCI (helix—the scientific explorer) 4843–4849, 2019) rotary kiln control techniques. In pyrometer-based measurements, the standard of radiation may lead to errors and inaccurate readings. Hence, the consuming zone temperature estimation got from the radiation pyrometer isn't solid and it is hard to get temperature data for a particular location. A colorimetric device-based intelligent control system measures the burning temperature of a specific point, but reading fluctuations are seen because of smoke and dust developed in the combustion process. In ROI based flame image processing, many factors, such as turbulent flame, brightness of flame zone, and dust, affect identifying the boundary for ROI based flame image analysis. In neural network models, variable selection plays a crucial role in designing effective systems with learning capabilities, but it is not an easy task to accomplish without certain rules. Hence, it is highly necessary to develop an improved control system. In view of the issues in variable and feature selection, a few neuro fuzzy systems are adopted in measurement and control. The consuming zone temperature estimation needs a lot of attention due to the very high temperature developed in the rotary kiln sintering process. Existing techniques have to be improved upon using advanced algorithms and intelligent approaches. A sintering state recognition system has been developed with features of flame images and fusion methodologies. In this approach, various flame image features and texture (Ren and Wang in Int J Autom Compu 11(1):72–77, 2014) features are extracted from the burning zone region. Though these methods address a few issues in flame image processing, the acquired image is largely affected by blurring and internal parameters of the sintering process. Charge coupled device (CCD) camera images and videos are applied to many image processing algorithms for better feature extraction and region extraction. The region of interest-based analysis is mainly focused on temperature assessment in this work. Intelligent control techniques are applied to measure the burning zone temperature in a rotary kiln. Fuzzy logic-based inference systems are combined with neural network algorithms in the development of neuro-fuzzy systems. The fuzzy surmising framework in light of mathematical models is the successful manner to anticipate the temperature esteems utilizing power measures. The fire pictures caught by the CCD cameras are handled utilizing fluffy rule-based picture investigation, which estimates temperature from a fire picture by contemplating RGB power planes. The arrangement of result temperature esteems is wanted to be a participation work. The Mamdani fluffy induction model is used to give planning of fluffy fire temperature. Exact temperature planning of fire pictures is performed to control the temperature inside the going stove to make top notch clinker. The fire picture examination is completed in different edges of three unique datasets, and temperature is estimated for various crude supper feed rates and coal feed rates. However there is a slight distinction in the acquired temperature, the general temperature evaluation process doesn't show a huge contrast as per the dataset.