Improving classification accuracy of fine-tuned CNN models: Impact of hyperparameter optimization.

The immense popularity of convolutional neural network (CNN) models has sparked a growing interest in optimizing their hyperparameters. Discovering the ideal values for hyperparameters to achieve optimal CNN training is a complex and time-consuming task, often requiring repetitive numerical experiments. As a result, significant attention is currently being devoted to developing methods aimed at tailoring hyperparameters for specific CNN models and classification tasks. While existing optimization methods often yield favorable image classification results, they do not provide guidance on which hyperparameters are worth optimizing, the appropriate value ranges for those hyperparameters, or whether it is reasonable to use a subset of training data for the optimization process. This work is focused on the optimization of hyperparameters during transfer learning, with the goal of investigating how different optimization methods and hyperparameter selections impact the performance of fine-tuned models. In our experiments, we assessed the importance of various hyperparameters and identified the ranges within which optimal CNN training can be achieved. Additionally, we compared four hyperparameter optimization methods-grid search, random search, Bayesian optimization, and the Asynchronous Successive Halving Algorithm (ASHA). We also explored the feasibility of fine-tuning hyperparameters using a subset of the training data. By optimizing the hyperparameters, we observed an improvement in CNN classification accuracy of up to 6%. Furthermore, we found that achieving a balance in class distribution within the subset of data used for parameter optimization is crucial in establishing the optimal set of hyperparameters for CNN training. The results we obtained demonstrate that hyperparameter optimization is highly dependent on the specific task and dataset at hand.

Detection of Fungi and Oomycetes by Volatiles Using E-Nose and SPME-GC/MS Platforms.

Fungi and oomycetes release volatiles into their environment which could be used for olfactory detection and identification of these organisms by electronic-nose (e-nose). The aim of this study was to survey volatile compound emission using an e-nose device and to identify released molecules through solid phase microextraction-gas chromatography/mass spectrometry (SPME-GC/MS) analysis to ultimately develop a detection system for fungi and fungi-like organisms. To this end, cultures of eight fungi (Armillaria gallica, Armillaria ostoyae, Fusarium avenaceum, Fusarium culmorum, Fusarium oxysporum, Fusarium poae, Rhizoctonia solani, Trichoderma asperellum) and four oomycetes (Phytophthora cactorum, P. cinnamomi, P. plurivora, P. ramorum) were tested with the e-nose system and investigated by means of SPME-GC/MS. Strains of F. poae, R. solani and T. asperellum appeared to be the most odoriferous. All investigated fungal species (except R. solani) produced sesquiterpenes in variable amounts, in contrast to the tested oomycetes strains. Other molecules such as aliphatic hydrocarbons, alcohols, aldehydes, esters and benzene derivatives were found in all samples. The results suggested that the major differences between respective VOC emission ranges of the tested species lie in sesquiterpene production, with fungi emitting some while oomycetes released none or smaller amounts of such molecules. Our e-nose system could discriminate between the odors emitted by P. ramorum, F. poae, T. asperellum and R. solani, which accounted for over 88% of the PCA variance. These preliminary results of fungal and oomycete detection make the e-nose device suitable for further sensor design as a potential tool for forest managers, other plant managers, as well as regulatory agencies such as quarantine services.

Odor Detection Using an E-Nose With a Reduced Sensor Array.

Recent advances in the field of electronic noses (e-noses) have led to new developments in both sensors and feature extraction as well as data processing techniques, providing an increased amount of information. Therefore, feature selection has become essential in the development of e-nose applications. Sophisticated computation techniques can be applied for solving the old problem of sensor number optimization and feature selections. In this way, one can find an optimal application-specific sensor array and reduce the potential cost associated with designing new e-nose devices. In this paper, we examine a procedure to extract and select modeling features for optimal e-nose performance. The usefulness of this approach is demonstrated in detail. We calculated the model's performance using cross-validation with the standard leave-one-group-out and group shuffle validation methods. Our analysis of wine spoilage data from the sensor array shows when a transient sensor response is considered, both from gas adsorption and desorption phases, it is possible to obtain a reasonable level of odor detection even with data coming from a single sensor. This requires adequate extraction of modeling features and then selection of features used in the final model.

Neural system for heartbeats recognition using genetically integrated ensemble of classifiers.

This paper presents the application of genetic algorithm for the integration of neural classifiers combined in the ensemble for the accurate recognition of heartbeat types on the basis of ECG registration. The idea presented in this paper is that using many classifiers arranged in the form of ensemble leads to the increased accuracy of the recognition. In such ensemble the important problem is the integration of all classifiers into one effective classification system. This paper proposes the use of genetic algorithm. It was shown that application of the genetic algorithm is very efficient and allows to reduce significantly the total error of heartbeat recognition. This was confirmed by the numerical experiments performed on the MIT BIH Arrhythmia Database.

May sunshine protect women against paroxysms of atrial fibrillation?

Seasonal variation in the occurrence of atrial fibrillation (AF) has been documented, yet precise mechanisms and factors underlying the phenomenon remain unclear. We have previously observed the decrease in the number of AF paroxysms between May and August, when sunshine levels were highest. The objective of the present study was, in turn, to determine whether sunshine affects the incidence of AF episodes. Participants were 1,475 Caucasian subjects (mean age: 68.2 years) diagnosed with AF paroxysms, admitted to the Intensive Cardiac Care Unit (ICCU) between January 1, 2005 and December 31, 2008; 805 were women and 670 were men (mean age: 69.2 and 66.2, respectively). The average incidence of AF episodes was higher among female subjects, with 16.8 cases per month, compared to male subjects with 14.0 cases per month. Pearson's correlation coefficient (r) was used to find a relationship between monthly sums of sunshine duration and AF paroxysms. This relationship for women was clearly inversely proportional (r = -0.499); namely, most AF episodes were recorded from December to March, when sunshine levels were lowest. In contrast, there was no noticeable association in male patients between the occurrence of AF paroxysms and effective sunshine (r = -0.126). In conclusion, unlike in men, a marked, statistically confirmed relationship between AF episodes and effective sunshine was observed in women. Thus, sunshine may have a protective effect against AF paroxysms for women. Our findings may provide the basic information concerning the influence of environmental factors on human wellbeing and contribute to management of AF.

Episodes of atrial fibrillation and meteorological conditions.

BACKGROUND: Atrial fibrillation (AF) is the most common arrhythmia encountered in clinical practice. The natural history of AF tends to begin with short paroxysms which gradually evolve into longer episodes, frequently treatment-resistant, and finally take a permanent form. It is a polyaetiological condition and single paroxysms may be caused by a variety of factors. There is a prevailing belief that weather is a vital element affecting the functioning of the human organism. Accordingly, high variability in hospital admissions due to AF paroxysms may be associated with meteorological conditions. AIM: To investigate the relationship between the incidence of AF paroxysms and atmospheric phenomena. METHODS: A total of 739 patients participated in the study [52% females, aged 18-91 (mean=65 years)], hospitalised for AF paroxysms in the Cardiac Care Unit (CCU) in 2005-2006. Patients with AF secondary to acute coronary syndrome, recent myocardial infarction, myocarditis, pericarditis, thyrotoxicosis, and disorders of the respiratory system, were excluded from the analysis. Statistical relationships were sought between the frequency of AF paroxysms and meteorological elements, such as: temperature change, atmospheric pressure, relative humidity, cloudiness, and wind speed. Using synoptic maps, such phenomena as weather fronts occurrence and baric systems were analysed. RESULTS: A considerable influence of a cold front and occlusion of cold front type on increases in admissions to CCU for AF paroxysms was observed. The absence of arrhythmia for many consecutive days was noted during the presence of stationary high-pressure areas. There were no significant relationships between meteorological elements and AF paroxysms. A seasonal distribution of AF episodes was found, with the maximum incidence in winter months and a decrease in the number of patients hospitalised from May to August. The impact of cold fronts may be explained by the effect of electromagnetic waves occurring in the zone of atmospheric changes, which may penetrate into buildings. On account of the translocation speed of electromagnetic waves, the effects may be felt many hours before an atmospheric front approaches. CONCLUSIONS: Meteorological conditions may have some influence on the occurrence of paroxysms of atrial fibrillation. This study could serve as a starting point for further research investigating relationships between weather conditions and heart rhythm disorders.