Develop a new method for the identification and identification of bacterial colonies using neural networks and Machine Learning (Machine Learning) algorithms. (Q77597): Difference between revisions
Jump to navigation
Jump to search
(Removed claim: summary (P836): Reference_reference_programme_aids:SA.41471 (2015/X) _public:Article 25 of Commission Regulation (EC) No 651/2014 of 17 June 2014 declaring certain categories of aid compatible with the internal market in the application of Article 107 and 108 of the Treaty (OJ(OJ LEU L 187/1, 26.06.2014).The aim of the project is to develop algorithms to automate the analysis of bacterial growth in Petri dishes, in particular the classification of the types o...) |
(Created claim: summary (P836): Reference number of the aid programme: SA.41471(2015/X) Purpose of public aid: Article 25 of EC Regulation No 651/2014 of 17 June 2014 declaring certain types of aid compatible with the internal market in the application of Articles 107 and 108 of the Treaty (OJ L. I'm sorry. EU L 187/1 of 26.06.2014). The aim of the project is to develop algorithms to automate the analysis of bacterial cultures on petri dishes, in particular the classification...) |
||||||||||||||
| Property / summary | |||||||||||||||
Reference number of the aid programme: SA.41471(2015/X) Purpose of public aid: Article 25 of EC Regulation No 651/2014 of 17 June 2014 declaring certain types of aid compatible with the internal market in the application of Articles 107 and 108 of the Treaty (OJ L. I'm sorry. EU L 187/1 of 26.06.2014). The aim of the project is to develop algorithms to automate the analysis of bacterial cultures on petri dishes, in particular the classification of the type of sown bacteria and the number of bacterial colonies. Algorithms will be based on deep learning methods, which, according to current scientific knowledge, are best suited for image recognition and can naturally be used to identify key morphological features of bacterial colonies. In cooperation with an expert in the field of microbiology, a training database will be prepared, needed to match the parameters of artificial neural networks. It will contain well-defined images of Petri dishes, on which well-defined bacterial strains have been planted and which can be used in supervised learning. Additional information on the samples will be provided by spectrum analysis (spectral) and a three-dimensional map of vertices made by laser scanning. Convolutional neural networks (convolutional neural networks) will be used to classify samples, whose hyperparameters will be selected for the problem under consideration and their optimal values found by cross-checking will avoid problems with " Underfitting or " Overfitting model. A comparative analysis of the results obtained with or without additional spectral data shall be performed. The final product will be specialised software using created algorithms to analyse bacterial cultures. (English) | |||||||||||||||
| Property / summary: Reference number of the aid programme: SA.41471(2015/X) Purpose of public aid: Article 25 of EC Regulation No 651/2014 of 17 June 2014 declaring certain types of aid compatible with the internal market in the application of Articles 107 and 108 of the Treaty (OJ L. I'm sorry. EU L 187/1 of 26.06.2014). The aim of the project is to develop algorithms to automate the analysis of bacterial cultures on petri dishes, in particular the classification of the type of sown bacteria and the number of bacterial colonies. Algorithms will be based on deep learning methods, which, according to current scientific knowledge, are best suited for image recognition and can naturally be used to identify key morphological features of bacterial colonies. In cooperation with an expert in the field of microbiology, a training database will be prepared, needed to match the parameters of artificial neural networks. It will contain well-defined images of Petri dishes, on which well-defined bacterial strains have been planted and which can be used in supervised learning. Additional information on the samples will be provided by spectrum analysis (spectral) and a three-dimensional map of vertices made by laser scanning. Convolutional neural networks (convolutional neural networks) will be used to classify samples, whose hyperparameters will be selected for the problem under consideration and their optimal values found by cross-checking will avoid problems with " Underfitting or " Overfitting model. A comparative analysis of the results obtained with or without additional spectral data shall be performed. The final product will be specialised software using created algorithms to analyse bacterial cultures. (English) / rank | |||||||||||||||
Normal rank | |||||||||||||||
| Property / summary: Reference number of the aid programme: SA.41471(2015/X) Purpose of public aid: Article 25 of EC Regulation No 651/2014 of 17 June 2014 declaring certain types of aid compatible with the internal market in the application of Articles 107 and 108 of the Treaty (OJ L. I'm sorry. EU L 187/1 of 26.06.2014). The aim of the project is to develop algorithms to automate the analysis of bacterial cultures on petri dishes, in particular the classification of the type of sown bacteria and the number of bacterial colonies. Algorithms will be based on deep learning methods, which, according to current scientific knowledge, are best suited for image recognition and can naturally be used to identify key morphological features of bacterial colonies. In cooperation with an expert in the field of microbiology, a training database will be prepared, needed to match the parameters of artificial neural networks. It will contain well-defined images of Petri dishes, on which well-defined bacterial strains have been planted and which can be used in supervised learning. Additional information on the samples will be provided by spectrum analysis (spectral) and a three-dimensional map of vertices made by laser scanning. Convolutional neural networks (convolutional neural networks) will be used to classify samples, whose hyperparameters will be selected for the problem under consideration and their optimal values found by cross-checking will avoid problems with " Underfitting or " Overfitting model. A comparative analysis of the results obtained with or without additional spectral data shall be performed. The final product will be specialised software using created algorithms to analyse bacterial cultures. (English) / qualifier | |||||||||||||||
point in time: 14 October 2020
| |||||||||||||||
Revision as of 10:12, 14 October 2020
Project in Poland financed by DG Regio
| Language | Label | Description | Also known as |
|---|---|---|---|
| English | Develop a new method for the identification and identification of bacterial colonies using neural networks and Machine Learning (Machine Learning) algorithms. |
Project in Poland financed by DG Regio |
Statements
1,850,331.93 zloty
0 references
2,437,811.1 zloty
0 references
75.9 percent
0 references
1 July 2018
0 references
30 October 2020
0 references
NEUROSYS SPÓŁKA Z OGRANICZONĄ ODPOWIEDZIALNOŚCIĄ
0 references
51°7'34.7"N, 16°58'41.5"E
0 references
Numer_referencyjny_programu_pomocowego: SA.41471(2015/X) Przeznaczenie_pomocy_publicznej: art. 25 rozporządzenia KE nr 651/2014 z dnia 17 czerwca 2014 r. uznające niektóre rodzaje pomocy za zgodne z rynkiem wewnętrznym w stosowaniu art. 107 i 108 Traktatu (Dz. Urz. UE L 187/1 z 26.06.2014). Celem projektu jest opracowanie algorytmów umożliwiających zautomatyzowanie analizy posiewów hodowli bakteryjnych na szalkach Petriego, w szczególności klasyfikacji rodzaju posianych bakterii oraz określenie liczby kolonii bakteryjnych. Algorytmy zostaną przygotowane w oparciu o metody uczenia głębokiego (ang. deep learning), które wg aktualnej wiedzy naukowej najlepiej sprawdzają się przy rozpoznawaniu obrazów i w naturalny sposób mogą być wykorzystane do identyfikowania kluczowych cech morfologicznych kolonii bakteryjnych. We współpracy z ekspertem z zakresu mikrobiologii zostanie przygotowana baza danych treningowych, potrzebnych do dopasowania parametrów sztucznych sieci neuronowych. Zawierać będzie ona dokładnie opisane zdjęcia szalek Petriego, na których zostały wysiane ściśle określone szczepy bakteryjne, a które będzie można wykorzystać w procesie uczenia nadzorowanego (ang. supervised learning). Dodatkowych informacji na temat próbek dostarczy analiza widma (spektralna) oraz trójwymiarowa mapa wierzchołków wykonana metodą skanowania laserowego. Do klasyfikacji próbek zostaną wykorzystane konwolucyjne sieci neuronowe (ang. convolutional neural networks), których hiperparametry zostaną dobrane pod rozważany problem, a ich optymalne wartości znalezione metodą sprawdzianu krzyżowego, co pozwoli uniknąć problemów z "niedouczeniem" (ang. underfitting) lub "przeuczeniem" (ang. overfitting) modelu. Zostanie wykonana analiza porównawcza wyników otrzymanych z lub bez uwzględnienia dodatkowych danych spektralnych. Ostatecznym produktem będzie specjalistyczne oprogramowanie wykorzystujące stworzone algorytmy do analizy posiewów bakteryjnych. (Polish)
0 references
Reference number of the aid programme: SA.41471(2015/X) Purpose of public aid: Article 25 of EC Regulation No 651/2014 of 17 June 2014 declaring certain types of aid compatible with the internal market in the application of Articles 107 and 108 of the Treaty (OJ L. I'm sorry. EU L 187/1 of 26.06.2014). The aim of the project is to develop algorithms to automate the analysis of bacterial cultures on petri dishes, in particular the classification of the type of sown bacteria and the number of bacterial colonies. Algorithms will be based on deep learning methods, which, according to current scientific knowledge, are best suited for image recognition and can naturally be used to identify key morphological features of bacterial colonies. In cooperation with an expert in the field of microbiology, a training database will be prepared, needed to match the parameters of artificial neural networks. It will contain well-defined images of Petri dishes, on which well-defined bacterial strains have been planted and which can be used in supervised learning. Additional information on the samples will be provided by spectrum analysis (spectral) and a three-dimensional map of vertices made by laser scanning. Convolutional neural networks (convolutional neural networks) will be used to classify samples, whose hyperparameters will be selected for the problem under consideration and their optimal values found by cross-checking will avoid problems with " Underfitting or " Overfitting model. A comparative analysis of the results obtained with or without additional spectral data shall be performed. The final product will be specialised software using created algorithms to analyse bacterial cultures. (English)
14 October 2020
0 references
Identifiers
POIR.01.01.01-00-0040/18
0 references