Enterprise Activity Modeling in Walnut Sector in Ukraine

The main goal of this study was to provide a critical analysis of the oil and fat sub-complex for deep walnut processing, to determine and compare the profitability of enterprises’ activities under different business models for implementation in the agro-food value chain. The latter was considered as an important factor for the development of the domestic market of walnuts and export opportunities. Business modeling of the enterprise activity in the oil and fat sub-complex for deep walnut processing was carried out. The stages of production and marketing activities of the enterprise from the garden planting or the purchase of the processed raw materials to the sale of the processed raw materials and products obtained from walnut processing depending on the chosen business model were considered. A comparative analysis of profitability of the enterprise activity and absolute values of income and profitability indicators under various business models of the enterprise activity were shown. The most cost-effective business-model entailed the combination of walnut production and its processing, which will provide profitability of up to 4640.32% in the 20th year of the project implementation. The results of the given study are intended for the agricultural enterprises of central region of Ukraine.

Human activity recognition using wearable sensors, discriminant analysis, and long short-term memory-based neural structured learning

Healthcare using body sensor data has been getting huge research attentions by a wide range of researchers because of its good practical applications such as smart health care systems. For instance, smart wearable sensor-based behavior recognition system can observe elderly people in a smart eldercare environment to improve their lifestyle and can also help them by warning about forthcoming unprecedented events such as falls or other health risk, to prolong their independent life. Although there are many ways of using distinguished sensors to observe behavior of people, wearable sensors mostly provide reliable data in this regard to monitor the individual’s functionality and lifestyle. In this paper, we propose a body sensor-based activity modeling and recognition system using time-sequential information-based deep Neural Structured Learning (NSL), a promising deep learning algorithm. First, we obtain data from multiple wearable sensors while the subjects conduct several daily activities. Once the data is collected, the time-sequential information then go through some statistical feature processing. Furthermore, kernel-based discriminant analysis (KDA) is applied to see the better clustering of the features from different activity classes by minimizing inner-class scatterings while maximizing inter-class scatterings of the samples. The robust time-sequential features are then applied with Neural Structured Learning (NSL) based on Long Short-Term Memory (LSTM), for activity modeling. The proposed approach achieved around 99% recall rate on a public dataset. It is also compared to existing different conventional machine learning methods such as typical Deep Belief Network (DBN), Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN) where they yielded the maximum recall rate of 94%. Furthermore, a fast and efficient explainable Artificial Intelligence (XAI) algorithm, Local Interpretable Model-Agnostic Explanations (LIME) is used to explain and check the machine learning decisions. The robust activity recognition system can be adopted for understanding peoples' behavior in their daily life in different environments such as homes, clinics, and offices.

Using in silico methods to obtain bioactive peptides of whey

Целью данного исследования являлся скрининг биопептидов, высвобождаемых из белков молочной сыворотки с использованием базы данных BIOPEP. В задачи работы входили оценка сывороточных белков как потенциальных предшественников биоактивных пептидов с последующей оценкой их потенциальной биологической активности, моделирование ферментативного гидролиза и оценка полученного пептидного профиля. Объектами исследований являлись белки молочной сыворотки и ферментные препараты трипсин EC (3.4.21.4), химотрипсин EC (3.4.21.1) и алкалаза (EC 3.4.21.62). Методы исследований включали in silico анализ с использованием базы данных BIOPEP-UWM™. В результате исследований в бета-лактоглобулине была определена наибольшая частота встречаемости гипотензивных пептидов, ингибирующих ангиотензин-I-превращающий фермент (АПФ) (A=0,5528), и противодиабетических, ингибирующих дипептидилпептидазу IV (ДПП-IV) (A=0,6584), суммарная доля которых составила более 65 % среди всех потенциальных биопептидов. Наилучший результат при дальнейшем моделировании in silico гидролиза бета-лактоглобулина был получен с использованием трипсина, химотрипсина и алкалазы. Показано, что данные ферменты способствуют выделению пептидов с высокими значениями IC50: гипотензивные (VY [41-42] (IC50 = 7,1 мкМ), VF [80-81] (IC50 = 9,2 мкМ), VY [41-42] (IC50 = 7,1 мкМ), IIAEK [70-74] (IC50 = 63,7 мкМ), VR [122-123] (IC50 = 141мкМ) и противодиабетические (VL [91-92] (IC50 = 74 мкМ), IPAVF (IC50 = 44,7 мкМ), VR [122-123] (IC50 = 52,8 мкМ). Данные проведенного биоинформационного подхода определили условия для последующего воспроизведения на реальных пищевых белковых системах. The aim of this study was to screen for biopeptides released from whey proteins using the BIOPEP database. The tasks of the work included: assessment of whey proteins as potential precursors of bioactive peptides with subsequent assessment of their potential biological activity, modeling of enzymatic hydrolysis and assessment of the obtained peptide profile.The objects of the study were whey proteins and enzyme preparations trypsin EC (3.4.21.4), chymotrypsin EC (3.4.21.1) and Alcalase (EC 3.4.21.62). Research methods are included in silicoanalysis using the BIOPEP-UWM ™ database. As a result of studies on beta-lactoglobulin, the highest frequency of occurrence of antihypertensive peptides that inhibit angiotensin-I-converting enzyme (ACE) (A = 0.5528) and antidiabetic peptides that inhibit dipeptidyl peptidase IV (DPP-IV) (A = 0.6584) was obtained, the total share of which was more than 65 % among all potential biopeptides. The best result in further in silico modeling of beta-lactoglobulin hydrolysis was obtained using trypsin, chymotrypsin, and alkalase. These enzymes were shown to promote the release of peptides with high IC50 values: hypotensive (VY [41-42] (IC50 = 7.1 μM), VF [80-81] (IC50 = 9.2 μM), VY [41-42] (IC50 = 7.1 μM), IIAEK [70-74] (IC50 = 63.7 μM), VR [122-123] (IC50 = 141 μM) and antidiabetic (VL [91-92] (IC50 = 74 μM), IPAVF (IC50 = 44.7 μM), VR [122-123] (IC50 = 52.8 μM). The data of the bioinformatic approach carried out determined the conditions for subsequent reproduction on real food protein systems.

Human Activity Recognition Using Wearable Sensors, Discriminant Analysis, and Long Short-Term Memory-based Neural Structured Learning

Healthcare using body sensor data has been getting huge research attentions by a wide range of researchers because of its good practical applications such as smart health care systems. For instance, smart wearable sensor-based behavior recognition system can observe elderly people in a smart eldercare environment to improve their lifestyle and can also help them by warning about forthcoming unprecedented events such as falls or other health risk, to prolong their independent life. Although there are many ways of using distinguished sensors to observe behavior of people, wearable sensors mostly provide reliable data in this regard to monitor the individual’s functionality and lifestyle. In this paper, we propose a body sensor-based activity modeling and recognition system using time-sequential information-based deep Neural Structured Learning (NSL), a promising deep learning algorithm. First, we obtain data from multiple wearable sensors while the subjects conduct several daily activities. Once the data is collected, the time-sequential information then go through some statistical feature processing. Furthermore, kernel-based discriminant analysis (KDA) is applied to see the better clustering of the features from different activity classes by minimizing inner-class scatterings while maximizing inter-class scatterings of the samples. The robust time-sequential features are then applied with Neural Structured Learning (NSL) based on Long Short-Term Memory (LSTM), for activity modeling. The proposed approach achieved around 99% recall rate on a public dataset. It is also compared to existing different conventional machine learning methods such as typical Deep Belief Network (DBN), Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN) where they yielded the maximum recall rate of 94%. Furthermore, a fast and efficient explainable Artificial Intelligence (XAI) algorithm, Local Interpretable Model-Agnostic Explanations (LIME) is used to explain and check the machine learning decisions. The robust activity recognition system can be adopted for understanding peoples' behavior in their daily life in different environments such as homes, clinics, and offices.