Democritus University of Thrace (DUTH)
Ανάπτυξη ευφυών προτύπων και αντίστοιχων πληροφοριακών συστημάτων, εμπνευσμένων από βιολογικά αντίστοιχα, με στόχο την αξιολόγηση περιβαλλοντικών προβλημάτων και κινδύνων
Abstract
dc:descriptionThe disturbances in the Earth's biosphere and the natural environment originating from irrational human activities have created serious environmental problems which are equivalent to an extensive and prolonged ecological crisis. The disturbances in the Earth's biosphere and the natural environment originating from irrational human activities have created serious environmental problems, which is equivalent to an extensive and prolonged ecological crisis.Climate change is revealed clearly, by the increase in global average air and ocean temperatures, through the widespread melting of snow and ice and by the rising of the global average sea level. It is an additional pressure on ecosystems, resulting in very dangerous situations with negative effects on human health and it damages agriculture, forestry, energy, tourism and infrastructure in general.One of the most serious potential impacts of climate change lies in the complex interaction of species in relation to climatic survival factors, and particularly in the spread of invasive species, which constitute a serious and rapidly worsening threat to natural biodiversity of the native environment and generally a threat to flora, fauna, and even to the human population.The critical stage for the adoption and implementation of a specific policy to address the elimination, control or containment of the species, is the process of their recognition, given that these species are usually unknown to the new installation environment. The process of recognition and safe identification is an extremely difficult, complicated and dangerous process as neither the large differences in morphology, nor the significant similarities, reflect the affinity or not between biological organisms.The necessity of a comprehensive and absolutely valid identification of these species is more than evident in the cases of treatment programs design the identification process depends on a multitude of information required and the constant monitoring of the status quo.So, after recognizing climate change as the root cause of the very major environmental problems, the objective of this PhD thesis is the employment of intelligent analysis of phenomena that enhance climate change (such as air pollution). It is especially crucial to apply and combine methods (in a hybrid mode) that can predict or model the potential effects of climate change such as increasing the frequency of extreme weather phenomena, extensive forest fires and the spread of invasive species.Specifically, this thesis proposes the development of a Hybrid Combined Computational Intelligence Information System, which is based exclusively on innovative methods of Biologically Inspired Computational Intelligence. This information system considers the quality of the atmosphere and can predict the extent of air pollution, which is a catalyst and a major factor in maximizing the effects of climate change.It also can make timely and accurate prediction for the occurrence of extreme weather events, which are associated with anthropogenic climate change and in particular with the continuous increase in average temperatures and with the rainfall reduction due to global warming.Additionally, the system can perform an analysis of the factors that significantly affect the forest fires, including the increasing frequency and intensity favored by drought, prolonged periods of high temperatures and extreme weather events, namely the impacts of climate change.Also this system can conduct a study of the spatiotemporal conditions of occurrence of harmful invasive species, in relation to climatic factors of survival and to the overall impact of climate change on habitat structure and ecological function.It can therefore analyze the changes observed in the populations of the species, changes in the way of their spread and to identify a series of phenological and ecological changes identified in the wider natural living and breeding environment.Finally, another very important and innovative process of automated identification of invasive species, either with phenotypic processes or with analysis of the species DNA, is effectively achieved by the proposed scheme.Specifically, the phenotypic identification is achieved through computer vision and through Machine Hearing. The recognition by DNA analysis is accomplished by the identification of the genetic barcode material (DNA barcoding) of every species and through an innovative identification system which is based on environmental DNA analysis (Metagenomics).To demonstrate the validity and reliability of the methods used, specific case studies were performed, in particularly difficult scenarios of decision making and forecasting. The whole process yielded extremely high accuracy results, which reinforces the view for adoption and widespread use of Biologically Inspired intelligent digital information systems.
Degree
thesis:*- Grantor dc:publisher
- Democritus University of Thrace (DUTH)
- Year dc:date
- 2017
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Demertzis, Konstantinos
- Δεμερτζής, Κωνσταντίνος
Subjects
dc:subject × 48- Υπολογιστική νοημοσύνη
- Ευφυή υπολογιστικά συστήματα
- Μηχανική μάθηση βιολογικής έμπνευσης
- Ασαφής λογική
- Ασαφείς γνωστικοί χάρτες
- Τεχνητά νευρωνικά δίκτυα
- Νευρωνικά δίκτυα αιχμής
- Δίκτυα ακραίας μάθησης
- Δίκτυα βαθιάς μάθησης
- Βιολογικά εμπνευσμένες μέθοδοι βελτιστοποίησης
- Μηχανική όραση
- Μηχανική ακοή
- Ανάλυση DNA
- Περιβαλλοντικό DNA
- DNA ραβδοκώδικας
- Μεταγονιδιωματική
- Κλιματική αλλαγή
- Ακραία καιρικά φαινόμενα
- Ατμοσφαιρική ρύπανση
- Δασικές πυρκαγιές
- Χωροκατακτητικά είδη
- Computational intelligence
- Intelligent computer systems
- Bio-inspired machine learning
- Fuzzy logic
- Fuzzy cognitive maps
- Artificial neural networks
- Spiking neural networks
- Extreme learning machines
- Deep neural networks
- Bio-inspired optimization methods
- Machine vision
- Machine hearing
- DNA analysis
- Environmental DNA
- DNA barcoding
- Metagenomics
- Climate change
- Extreme weather events
- Air pollution
- Forest fires
- Invasive species
- Φυσικές Επιστήμες
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Γεωεπιστήμες και Επιστήμες Περιβάλλοντος
- Natural Sciences
- Computer and Information Sciences
- Earth and Related Environmental Sciences
Rights
dc:rights- Statement dc:rights
-
- BY
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/39958