Automatic Fungi Recognition: Deep Learning Meets Mycology

The article presents an AI-based fungi species recognition system for a citizen-science community. The system’s real-time identification too — FungiVision — with a mobile application front-end, led to increased public interest in fungi, quadrupling the number of citizens collecting data. FungiVision, deployed with a human-in-the-loop, reaches nearly 93% accuracy. Using the collected data, we developed a novel fine-grained classification dataset — Danish Fungi 2020 (DF20) — with several unique characteristics: species-level labels, a small number of errors, and rich observation metadata. The dataset enables the testing of the ability to improve classification using metadata, e.g., time, location, habitat and substrate, facilitates classifier calibration testing and finally allows the study of the impact of the device settings on the classification performance. The continual flow of labelled data supports improvements of the online recognition system. Finally, we present a novel method for the fungi recognition service, based on a Vision Transformer architecture. Trained on DF20 and exploiting available metadata, it achieves a recognition error that is 46.75% lower than the current system. By providing a stream of labeled data in one direction, and an accuracy increase in the other, the collaboration creates a virtuous cycle helping both communities.

Download Full-text

Investigating the Impact of Human in-the-Loop Digital Twin in an Industrial Maintenance Context

SSRN Electronic Journal ◽

10.2139/ssrn.3717797 ◽

2020 ◽

Author(s):

Ali Al-Yacoubb ◽

Will Eaton ◽

Melanie Zimmer ◽

Achim Buerkle ◽

Dedy Ariansyaha ◽

...

Keyword(s):

Digital Twin ◽

Human In The Loop ◽

Industrial Maintenance ◽

The Impact

Download Full-text

Hierarchical Phoneme Classification for Improved Speech Recognition

Applied Sciences ◽

10.3390/app11010428 ◽

2021 ◽

Vol 11 (1) ◽

pp. 428

Author(s):

Donghoon Oh ◽

Jeong-Sik Park ◽

Ji-Hwan Kim ◽

Gil-Jin Jang

Keyword(s):

Speech Recognition ◽

Language Processing ◽

Confusion Matrix ◽

Critical Factor ◽

Recognition System ◽

Classification Performance ◽

Language Models ◽

Successful Implementation ◽

Phoneme Classification ◽

Improved Performance

Speech recognition consists of converting input sound into a sequence of phonemes, then finding text for the input using language models. Therefore, phoneme classification performance is a critical factor for the successful implementation of a speech recognition system. However, correctly distinguishing phonemes with similar characteristics is still a challenging problem even for state-of-the-art classification methods, and the classification errors are hard to be recovered in the subsequent language processing steps. This paper proposes a hierarchical phoneme clustering method to exploit more suitable recognition models to different phonemes. The phonemes of the TIMIT database are carefully analyzed using a confusion matrix from a baseline speech recognition model. Using automatic phoneme clustering results, a set of phoneme classification models optimized for the generated phoneme groups is constructed and integrated into a hierarchical phoneme classification method. According to the results of a number of phoneme classification experiments, the proposed hierarchical phoneme group models improved performance over the baseline by 3%, 2.1%, 6.0%, and 2.2% for fricative, affricate, stop, and nasal sounds, respectively. The average accuracy was 69.5% and 71.7% for the baseline and proposed hierarchical models, showing a 2.2% overall improvement.

Download Full-text

Automated classification of clinical trial eligibility criteria text based on ensemble learning and metric learning

BMC Medical Informatics and Decision Making ◽

10.1186/s12911-021-01492-z ◽

2021 ◽

Vol 21 (S2) ◽

Author(s):

Kun Zeng ◽

Yibin Xu ◽

Ge Lin ◽

Likeng Liang ◽

Tianyong Hao

Keyword(s):

Clinical Trial ◽

Ensemble Learning ◽

Metric Learning ◽

Classification Performance ◽

Ensemble Model ◽

Automated Classification ◽

Eligibility Criteria ◽

Data Imbalance ◽

The Impact

Abstract Background Eligibility criteria are the primary strategy for screening the target participants of a clinical trial. Automated classification of clinical trial eligibility criteria text by using machine learning methods improves recruitment efficiency to reduce the cost of clinical research. However, existing methods suffer from poor classification performance due to the complexity and imbalance of eligibility criteria text data. Methods An ensemble learning-based model with metric learning is proposed for eligibility criteria classification. The model integrates a set of pre-trained models including Bidirectional Encoder Representations from Transformers (BERT), A Robustly Optimized BERT Pretraining Approach (RoBERTa), XLNet, Pre-training Text Encoders as Discriminators Rather Than Generators (ELECTRA), and Enhanced Representation through Knowledge Integration (ERNIE). Focal Loss is used as a loss function to address the data imbalance problem. Metric learning is employed to train the embedding of each base model for feature distinguish. Soft Voting is applied to achieve final classification of the ensemble model. The dataset is from the standard evaluation task 3 of 5th China Health Information Processing Conference containing 38,341 eligibility criteria text in 44 categories. Results Our ensemble method had an accuracy of 0.8497, a precision of 0.8229, and a recall of 0.8216 on the dataset. The macro F1-score was 0.8169, outperforming state-of-the-art baseline methods by 0.84% improvement on average. In addition, the performance improvement had a p-value of 2.152e-07 with a standard t-test, indicating that our model achieved a significant improvement. Conclusions A model for classifying eligibility criteria text of clinical trials based on multi-model ensemble learning and metric learning was proposed. The experiments demonstrated that the classification performance was improved by our ensemble model significantly. In addition, metric learning was able to improve word embedding representation and the focal loss reduced the impact of data imbalance to model performance.

Download Full-text

Reliability Assessment of a Multi-State HVDC System by Combining Markov and Matrix-Based Methods

Energies ◽

10.3390/en14113097 ◽

2021 ◽

Vol 14 (11) ◽

pp. 3097

Author(s):

Roberto Benato ◽

Antonio Chiarelli ◽

Sebastian Dambone Sessa

Keyword(s):

Current System ◽

Estimation Method ◽

Reliability Estimation ◽

Voltage Source ◽

Voltage Source Converter ◽

Hvdc System ◽

Critical Elements ◽

The Matrix ◽

The Impact

The purpose of this paper is to highlight that, in order to assess the availability of different HVDC cable transmission systems, a more detailed characterization of the cable management significantly affects the availability estimation since the cable represents one of the most critical elements of such systems. The analyzed case study consists of a multi-terminal direct current system based on both line commutated converter and voltage source converter technologies in different configurations, whose availability is computed for different transmitted power capacities. For these analyses, the matrix-based reliability estimation method is exploited together with the Monte Carlo approach and the Markov state space one. This paper shows how reliability analysis requires a deep knowledge of the real installation conditions. The impact of these conditions on the reliability evaluation and the involved benefits are also presented.

Download Full-text

Impact of Dataset Size on Classification Performance: An Empirical Evaluation in the Medical Domain

Applied Sciences ◽

10.3390/app11020796 ◽

2021 ◽

Vol 11 (2) ◽

pp. 796

Author(s):

Alhanoof Althnian ◽

Duaa AlSaeed ◽

Heyam Al-Baity ◽

Amani Samha ◽

Alanoud Bin Dris ◽

...

Keyword(s):

Empirical Evaluation ◽

Classification Performance ◽

Support Vector ◽

Robust Model ◽

Original Distribution ◽

C4.5 Decision Tree ◽

Dataset Size ◽

Overall Performance ◽

Medical Domain ◽

The Impact

Dataset size is considered a major concern in the medical domain, where lack of data is a common occurrence. This study aims to investigate the impact of dataset size on the overall performance of supervised classification models. We examined the performance of six widely-used models in the medical field, including support vector machine (SVM), neural networks (NN), C4.5 decision tree (DT), random forest (RF), adaboost (AB), and naïve Bayes (NB) on eighteen small medical UCI datasets. We further implemented three dataset size reduction scenarios on two large datasets and analyze the performance of the models when trained on each resulting dataset with respect to accuracy, precision, recall, f-score, specificity, and area under the ROC curve (AUC). Our results indicated that the overall performance of classifiers depend on how much a dataset represents the original distribution rather than its size. Moreover, we found that the most robust model for limited medical data is AB and NB, followed by SVM, and then RF and NN, while the least robust model is DT. Furthermore, an interesting observation is that a robust machine learning model to limited dataset does not necessary imply that it provides the best performance compared to other models.

Download Full-text

Potential Impacts of Future Climate Change Scenarios on Ground Subsidence

Water ◽

10.3390/w12010219 ◽

2020 ◽

Vol 12 (1) ◽

pp. 219 ◽

Cited By ~ 2

Author(s):

Antonio-Juan Collados-Lara ◽

David Pulido-Velazquez ◽

Rosa María Mateos ◽

Pablo Ezquerro

Keyword(s):

Climate Change ◽

Land Subsidence ◽

Ground Subsidence ◽

Lower Percentage ◽

Future Climate Change ◽

Climate Change Scenarios ◽

Fine Grained ◽

Fine Grained Material ◽

The Impact

In this work, we developed a new method to assess the impact of climate change (CC) scenarios on land subsidence related to groundwater level depletion in detrital aquifers. The main goal of this work was to propose a parsimonious approach that could be applied for any case study. We also evaluated the methodology in a case study, the Vega de Granada aquifer (southern Spain). Historical subsidence rates were estimated using remote sensing techniques (differential interferometric synthetic aperture radar, DInSAR). Local CC scenarios were generated by applying a bias correction approach. An equifeasible ensemble of the generated projections from different climatic models was also proposed. A simple water balance approach was applied to assess CC impacts on lumped global drawdowns due to future potential rainfall recharge and pumping. CC impacts were propagated to drawdowns within piezometers by applying the global delta change observed with the lumped assessment. Regression models were employed to estimate the impacts of these drawdowns in terms of land subsidence, as well as to analyze the influence of the fine-grained material in the aquifer. The results showed that a more linear behavior was observed for the cases with lower percentage of fine-grained material. The mean increase of the maximum subsidence rates in the considered wells for the future horizon (2016–2045) and the Representative Concentration Pathway (RCP) scenario 8.5 was 54%. The main advantage of the proposed method is its applicability in cases with limited information. It is also appropriate for the study of wide areas to identify potential hot spots where more exhaustive analyses should be performed. The method will allow sustainable adaptation strategies in vulnerable areas during drought-critical periods to be assessed.

Download Full-text

Novel Features and PRPD Image Denoising Method for Improved Single-Source Partial Discharges Classification in On-Line Hydro-Generators

Energies ◽

10.3390/en14113267 ◽

2021 ◽

Vol 14 (11) ◽

pp. 3267

Author(s):

Ramon C. F. Araújo ◽

Rodrigo M. S. de Oliveira ◽

Fernando S. Brasil ◽

Fabrício J. B. Barros

Keyword(s):

Image Denoising ◽

Recognition Performance ◽

Partial Discharge ◽

Classification Performance ◽

Partial Discharges ◽

Denoising Method ◽

Internal Parameters ◽

On Line ◽

Automatic Removal ◽

The Impact

In this paper, a novel image denoising algorithm and novel input features are proposed. The algorithm is applied to phase-resolved partial discharge (PRPD) diagrams with a single dominant partial discharge (PD) source, preparing them for automatic artificial-intelligence-based classification. It was designed to mitigate several sources of distortions often observed in PRPDs obtained from fully operational hydroelectric generators. The capabilities of the denoising algorithm are the automatic removal of sparse noise and the suppression of non-dominant discharges, including those due to crosstalk. The input features are functions of PD distributions along amplitude and phase, which are calculated in a novel way to mitigate random effects inherent to PD measurements. The impact of the proposed contributions was statistically evaluated and compared to classification performance obtained using formerly published approaches. Higher recognition rates and reduced variances were obtained using the proposed methods, statistically outperforming autonomous classification techniques seen in earlier works. The values of the algorithm’s internal parameters are also validated by comparing the recognition performance obtained with different parameter combinations. All typical PD sources described in hydro-generators PD standards are considered and can be automatically detected.

Download Full-text

Impact of UAS with Low Size, Weight, and Power Sensors on Air Traffic Controllers’ Performance and Acceptability Ratings

Proceedings of the Human Factors and Ergonomics Society Annual Meeting ◽

10.1177/1071181320641039 ◽

2020 ◽

Vol 64 (1) ◽

pp. 154-158

Author(s):

Kim-Phuong L. Vu ◽

Jonathan VanLuven ◽

Timothy Diep ◽

Vernol Battiste ◽

Summer Brandt ◽

...

Keyword(s):

Air Traffic ◽

Unmanned Aircraft ◽

Unmanned Aircraft Systems ◽

Subjective Ratings ◽

Detection Range ◽

Human In The Loop ◽

Air Traffic Controllers ◽

Low Altitude ◽

Sector Performance ◽

The Impact

A human-in-the-loop simulation was conducted to evaluate the impact of Unmanned Aircraft Systems (UAS) with low size, weight, and power (SWaP) sensors operating in a busy, low-altitude sector. Use of low SWaP sensors allow for UAS to perform detect-and-avoid (DAA) maneuvers against non-transponding traffic in the sector. Depending upon the detection range of the low SWaP sensor, the UAS pilot may or may not have time to coordinate with air traffic controllers (ATCos) prior to performing the DAA maneuver. ATCo’s sector performance and subjective ratings of acceptability were obtained in four conditions that varied in UAS-ATCo coordination (all or none) prior to the DAA maneuver and workload (higher or lower). For performance, ATCos committed more losses of separation in high than low workload conditions. They also had to make more flight plan changes to manage the UAS when the UAS pilot did not coordinate DAA maneuvers compared to when they did coordinate the maneuvers prior to execution. Although the ATCos found the DAA procedures used by the UAS in the study to be acceptable, most preferred the UAS pilot to coordinate their DAA maneuvers with ATCos prior to executing them.

Download Full-text

A Study of the Impact of Graphite on the Kinetics of SPS in Nano- and Submicron WC-10%Co Powder Compositions

Ceramics ◽

10.3390/ceramics4020025 ◽

2021 ◽

Vol 4 (2) ◽

pp. 331-363

Author(s):

Eugeniy Lantcev ◽

Aleksey Nokhrin ◽

Nataliya Malekhonova ◽

Maksim Boldin ◽

Vladimir Chuvil'deev ◽

...

Keyword(s):

Activation Energy ◽

Solid Phase ◽

Continuous Heating ◽

Diffusion Creep ◽

Hard Alloys ◽

Fine Grained ◽

Plasma Sintering ◽

Spark Plasma ◽

The Impact ◽

Kinetics Of

This study investigates the impact of carbon on the kinetics of the spark plasma sintering (SPS) of nano- and submicron powders WC-10wt.%Co. Carbon, in the form of graphite, was introduced into powders by mixing. The activation energy of solid-phase sintering was determined for the conditions of isothermal and continuous heating. It has been demonstrated that increasing the carbon content leads to a decrease in the fraction of η-phase particles and a shift of the shrinkage curve towards lower heating temperatures. It has been established that increasing the graphite content in nano- and submicron powders has no significant effect on the SPS activation energy for “mid-range” heating temperatures, QS(I). The value of QS(I) is close to the activation energy of grain-boundary diffusion in cobalt. It has been demonstrated that increasing the content of graphite leads to a significant decrease in the SPS activation energy, QS(II), for “higher-range” heating temperatures due to lower concentration of tungsten atoms in cobalt-based γ-phase. It has been established that the sintering kinetics of fine-grained WC-Co hard alloys is limited by the intensity of diffusion creep of cobalt (Coble creep).

Download Full-text

Trust, efficacy and ethicacy when testing prisoners for COVID-19

International Journal of Prisoner Health ◽

10.1108/ijph-10-2020-0084 ◽

2021 ◽

Vol ahead-of-print (ahead-of-print) ◽

Author(s):

Steve Lambert ◽

Dean Wilkinson

Keyword(s):

Large Scale ◽

Current System ◽

Epidemiological Studies ◽

Careful Consideration ◽

Power Relationships ◽

England And Wales ◽

Content Type ◽

Large Scale Testing ◽

The Uk ◽

The Impact

Purpose The outbreak of the severe acute respiratory syndrome coronavirus 2 virus and subsequent COVID-19 illness has had a major impact on all levels of society internationally. The extent of the impact of COVID-19 on prison staff and prisoners in England and Wales is unknown. Testing for COVID-19 both asymptomatic and symptomatic, as well as for antibodies, to date, has been minimal. The purpose of this paper is to explore the widespread testing of COVID-19 in prisons poses philosophical and ethical questions around trust, efficacy and ethicacy. Design/methodology/approach This paper is both descriptive, providing an overview of the widespread testing of COVID-19 in prisoners in England and Wales, and conceptual in that it discusses and argues the issues associated with large-scale testing. This paper provides a discussion, using comparative studies, of the issues associated with large-scale testing of prisoners across the prison estate in England and Wales (120 prisons). The issues identified in this paper are contextualised through the lens of COVID-19, but they are equally transferrable to epidemiological studies of any pandemic. Given the prevalence of COVID-19 globally and the lack of information about its spread in prisons, at the time of writing this paper, there is a programme of asymptomatic testing of prisoners. However, there remains a paucity of data on the spread of COVID-19 in prisons because of the progress with the ongoing testing programme. Findings The authors argue that the widespread testing of prisoners requires careful consideration of the details regarding who is included in testing, how consent is gained and how tests are administered. This paper outlines and argues the importance of considering the complex nuance of power relationships within the prison system, among prisoner officers, medical staff and prisoners and the detrimental consequences. Practical implications The widespread testing of COVID-19 presents ethical and practical challenges. Careful planning is required when considering the ethics of who should be included in COVID-19 testing, how consent will be gained, who and how tests will be administered and very practical challenges around the recording and assigning of COVID-19 test kits inside the prison. The current system for the general population requires scanning of barcodes and registration using a mobile number; these facilities are not permitted inside a prison. Originality/value This paper looks at the issues associated with mass testing of prisoners for COVID-19. According to the authors’ knowledge, there has not been any research that looks at the issues of testing either in the UK or internationally. The literature available details countries’ responses to the pandemic rather and scientific papers on the development of vaccines. Therefore, this paper is an original review of some of the practicalities that need to be addressed to ensure that testing can be as successful as possible.

Download Full-text