Soft computing is a branch of computer science that deals with system solutions based on artificial intelligence and natural selection. Soft computing techniques are able to handle imprecision, uncertainty, partial truth, and approximation, which are often encountered in real-life problems. Soft computing techniques are also adaptive, learning from data and experience, and can cope with dynamic and complex environments.
Soft computing includes various methods, such as:
Genetic algorithm (GA): A population-based optimization method that mimics the process of natural evolution. GA works by generating and evaluating candidate solutions using operators such as selection, crossover, and mutation.
Neural network (NN): A computational model that simulates the structure and function of biological neurons. NN works by processing information through layers of interconnected nodes that can learn from data and adjust their weights accordingly.
Fuzzy logic (FL): A logic system that allows for degrees of truth rather than binary values. FL works by using fuzzy sets and rules to represent and reason with vague and uncertain information.
Evolutionary algorithms (EA): A type of meta-heuristic algorithm that are inspired by the process of natural evolution. They work by maintaining a population of candidate solutions that undergo selection, crossover, and mutation operators to generate new solutions. Some examples of evolutionary algorithms are differential evolution, particle swarm optimization, and ant colony optimization.
Soft computing has many applications in different domains of computer science, such as:
Timetabling: The problem of assigning tasks or events to time slots and resources, such as classes to rooms and teachers. Soft computing can help to find feasible and optimal timetables that satisfy various constraints and preferences.
Location problems: The problem of finding the best location for a facility or service, such as a warehouse, a hospital, or a fire station. Soft computing can help to find optimal locations that minimize costs, distances, or risks, while considering various factors and uncertainties.
Here, we will share some of our previous works in the area of soft computing and its applications. We will show how soft computing can help us solve various computer science problems that are challenging and complex. We hope you will find these works interesting and useful. Since a separate post is considered for neural networks and deep learning, we omitted some of our NN works from here.
Today, the air pollution is a serious environmental problem becoming global concern for human beings Air quality is influenced by emissions, meteorological parameters and topography. The effect of these parameters can be predicted using statistical methods. In current study, the data in the period of March 2012 to October 2013 are used. These data have been gathered from the stations of the Department of Environment and Air Quality Control Organization (Azadi and Sharif stations) in Tehran city. The main purpose was to predict the air quality of the next day and emissions of carbon monoxide and suspended particles under the influence of instability indices and meteorological parameters using the Artificial Neural Network. Results of modeling process showed that the concentration of pollutants is strongly influenced by meteorological parameters. In addition, prediction of the PM10 concentration of the next day using meteorological parameters (RMSE=29.03, R=0.76), instability indices and meteorological parameters (RMSE=28.13, R=0.76) were better than those obtained for AQI predicted by meteorological parameters (RMSE=20.81, R=0.50) and instability indices and meteorological parameters (RMSE=19.23, R=0.47). In general, the predicted values of PM10 and CO were better compared to AQI. It can be concluded that artificial neural network couldn’t load the model properly for AQI compared to PM10.
ارزیابی دقت شبکه های عصبی مصنوعی (MLP و RBF) در پیش بینی گرد و غبار کارخانه سیمان سبزوار
Estimation of Artificial Neural Networks (MLP and RBF) Accuracy in Anticipation of the Dust of the Sabzevar Cement Factory, Journal of Research in Environmental Health)
Sensitivity Analysis of Meteorological Parameters and Instability Indices on Concentration of Carbon Monoxide, Particulate Matter, and Air Quality Index in Tehran
R. Farhadi, M. Hadavifar, M. Moeinaddini, and 1 more author
Aims: Nowadays, dangerous chemical pollutants by a numerous of natural and synthetic sources are produced and released to the environment. These pollutants have short-term and long-term effects on human health. The purpose of this paper is to examine the impact of climate parameters and instability indices on air pollution in Tehran-Iran. Materials and Methods: To evaluate the impact of meteorological parameters and indices of stability and instability on sensitivity analysis in Tehran-Iran, the Sharif University monitoring station was selected for air sampling and analysis. Sampling was performed from March 2011 to July 2012 in Tehran. Findings: Results of sensitivity analysis showed that average daily change of the concentration of pollutants throughout the year was very different and intensively influenced by meteorological parameters. Results showed that wind direction (WD) (82%) and relative humidity (32%) and temperature (20%) have the most influence on the concentration values of pollutants carbon monoxide (CO), particulate matter (PM10), and air quality index (AQI). The highest concentrations of CO occurred in summer and lowest in winter, and maximum concentration of PM10 was in autumn, and its lowest concentration was in spring. Results revealed that the lowest average of AQI occurred in the spring, while in autumn, winter, and summer have almost equal values, but in winter AQI has slightly higher values. Conclusion: According to the results of this research in Sharif station Tehran, the WD has the highest impact percentage (82%) on the concentration of pollutants. The highest concentrations of CO occurred in summer, and maximum concentration of PM10 was in autumn.
2017
ارزیابی عملکرد ماشین یادگیر نهایی در قطعهبندی تصاویر
در مسئله برش مینیمم هدف مینیمم کردن ظرفیت یالهای برش است. از روشهای تقریبی حل این مسائل میتوان به الگوریتم کارگِر اشاره کرد. که از تلفیق لبه ها به صورت تصادفی استفاده میکند .در این مقاله از شبیهسازی تبریدی برای حل این مسئله استفاده شده است و نتایج آن با روش کارگِر مقایسه شده است. نتایج آزمایشات برتری روش پیشنهادی را نسبت به روش کارگِر از منظر سرعت اجرا، نرخ همگرایی و میانگین خطا نشان داده است.
در مسئله برش مینیمم هدف مینیمم کردن ظرفیت یالهای برش است. از روشهای تقریبی حل این مسائل میتوان به الگوریتم کارگِر اشاره کرد. که از تلفیق لبه ها به صورت تصادفی استفاده میکند .در این مقاله از جستجوی ممنوعه برای حل این مسئله استفاده شده است و نتایج آن با روش کارگِر مقایسه شده است. نتایج آزمایشات برتری روش پیشنهادی را نسبت به روش کارگِر از منظر سرعت اجرا، نرخ همگرایی و میانگین خطا نشان داده است.
مسئله مکانیابی p -هاب با ظرفیت نامتناهی در حضور صف M/G/1
مسئله مکانیابی هاب یک تعمیم نسبتاً جدید از مسائل مکانیابی است. این مسائل با پیدا کردن مکانهای هاب و تخصیص نقاط تقاضا به این مکانها سرو کار دارد.ما هابها را که بخشهای پر ازدحام شبکه هستند، همانند یک صف M/G/1 مدلبندی میکنیم. در این مقاله ابتدا یک برنامهریزی غیر خطی با محدودیتهای خطی برای مسئله نمایش میدهیم که زمان کلی حمل و نقل بین گرههای شبکه را مینیمم میکند، سپس این مسئله را با استفاده از الگوریتم ژنتیک حل میکنیم و با الگوریتم جستجوی ممنوعه مقایسه میکنیم.
2013
تشخیص ناحیه چربی در تصاویر MRI با استفاده از شبكه عصبی با كوپلاژ پالسی
شناسایی ناحیه مرتبط با یک بافت خاص اهمیت زیادی در پزشکی و فیزیولوژی دارد. در این مقاله راهكاری برای تشخیص و جداسازی ناحیه چربی در تصاویر MRI ران پا مبتنی بر شبکه عصبی با کوپلاژ پالسی ارائه شده است . هدف اصلی سنجش میزان تاثیر ورزشهای خاص در کاهش یا افزایش حجم چربی ران بوده است. الگوریتمهای متفاوتی برای این كار پیادهسازی و مورد استفاده قرار گرفت. نتایج آزمایشات انجام شده نشان داد که برای این کاربرد خاص، شبكه عصبی با كوپلاژ پالسی، بهترین نتیجه قطعهبندی را بدست میدهد.
2007
A Fish School Clustering Algorithm: Applied to Student Sectioning Problem
M. Amintoosi, M. Fathy, N. Mozayani, and 1 more author
Dynamics of Continuous Discrete & Impulse Systems, series B: Applications and Algorithms, Dec 2007
Post Proceeding of LSMS2007, Life System Modeling and Simulation 2007, China