Monte Carlo Concrete DropPath for epistemic uncertainty estimation in pollen images classification
Annotation
The paper presents the results of a new method for training the NASNet neural network called Monte Carlo Concrete DropPath for epistemic uncertainty estimation to classify pollen images. The developed method is compared with existing methods for epistemic uncertainty estimation. The method turns an arbitrary multipath neural network into a Bayesian one by sampling from the predictive distribution. The Monte Carlo method samples different masks of DropPath to estimate uncertainty. Moreover, the probability of DropPath is optimized using continuous relaxation. The proposed method was tested for the classification task on the state-of-the-art NASNet architecture. The method demonstrated advantages on the task of classifying pollen images. The classification accuracy increased for 13 pollen species of allergen plants by 0.73 % on average compared to the baseline NASNet, reaching 98.34 % by F1 measure. Furthermore, the method increased calibration and reduced the epistemic uncertainty of the model by two times compared to the NASNets ensemble. It is shown that continuous relaxation of the DropPath probability parameter increases the accuracy of problem solving and reduces the epistemic uncertainty of the model. These results contribute to the automation of aeropalinological monitoring to reduce the time of informing patients who suffer from pollinosis and hence to prevent allergy symptoms. The developed method can be applied to train a neural network for other computer vision tasks on any image dataset.
Keywords
Постоянный URL
Articles in current issue
- A fiber optic vibration sensor based on SMF-MMF-SMF structure and a tilted fiber Bragg grating
- Aberration analysis of a wedge as a compensator element in augmented and virtual reality systems.
- Estimation of temperature detection delay in a fiber optic gyroscope sensing coil.
- The relative intensity noise of a vertical-cavity surface-emitting laser with a fiber Bragg grating external cavity
- A comparative analysis of the frequency spectra of the human eye and a cut diamond in visible light
- Vulnerabilities in the quantum key distribution system induced under a pulsed laser attack
- Proactive management of the composition and structure of the spatial monitoring system under the influence of destabilizing factors
- An algorithm of trajectory control for the movement of a mobile robot without measuring the position coordinates
- Different III-V semiconductor nanowires with quantum dots on silicon: growth by molecular-beam epitaxy and properties
- Impact of magnesium oxide concentration and yttrium-aluminum garnet stoichiometry deviation on the microstructure and optical transmission of YAG-based ceramics
- Measurements of heat capacity and thermal conductivity of β-Ga2O3 and β-(AlxGa1–x)2O3 bulk crystals grown by the Czochralski method
- An approach to the identification of the state of elements in cyber-physical systems based on principal component analysis
- Stochastic software testing for vulnerability analysis
- Research of machine learning methods in the problem of identification of blood cells
- An Enhanced Exploration and Exploitation of Modified Grey Wolf Optimizer for Fuzzy Rules Reduction in Cloud Intrusion Detection System (CIDS)
- Model of information interaction channel routing on the FANET network using fuzzy logic
- Bi-level authentication and mannequin routing for improving security features of WSN-IoT
- Spline-wavelet bent robust codes
- Social media user identity linkage by graphic content comparison
- Random number generation with arrays of coupled quantum-dot micropillar lasers
- The method of forming and using a digital passport for an electronic product at enterprises of the instrument-making industry
- Methods for determining the phase difference of the output signals for the receiving elements of a phased antenna array in a radiophoton circuit with parallel and series pairwise connection ...
- Routing in networks of autonomous underwater vehicles