Journal: Animals : an Open Access Journal from MDPI

Article Title: Efficient Convolutional Neural Network Model for the Taxonomy and Sex Identification of Three Phlebotomine Sandfly Species (Diptera, Psychodidae, and Phlebotominae)

doi: 10.3390/ani14243712

Figure Lengend Snippet: Details of the baseline CNN model layers (in order from input). S: Spatial, C: Channel, and B: Batch observations.

Article Snippet: The base CNN model and its variations were implemented using Matlab R2024a.

Techniques:

Journal: Frontiers in Neuroscience

Article Title: A Deep Network Model on Dynamic Functional Connectivity With Applications to Gender Classification and Intelligence Prediction

doi: 10.3389/fnins.2020.00881

Figure Lengend Snippet: ROC and accuracy for the gender classification task. (A) ROC curve across 10 testing folds for the CNN + LSTM and the SVM model. (B) Gender classification accuracy averaged across 10 testing folds. Bars refer to mean accuracy of all testing folds. Error bars indicate the standard error. Obviously, the CNN + LSTM model is statistically better than the SVM model (*** p < 0.001). (C) Accuracies of gender classification using the CNN + LSTM model on the real BOLD signals and their surrogate copies. A total of 100 surrogate data were generated by using MVPR to estimate the null distribution of classification accuracies (see section “Materials and Methods” for detail). With the mean classification accuracies as the statistic, results reveal that the classifier learned the connection dynamics with a probability of being wrong of <0.001. (D) The learning curves while training the CNN + LSTM model.

Article Snippet: For comparing with the CNN + LSTM model, we used linear SVM and linear epsilon SVR models (LIBSVM toolbox in Matlab ) based on dynamic characteristic of dFC (dFC-Str, which refers to the overall strength of dFC) , to achieve gender classification and intelligence prediction with the same 10-fold cross-validation strategies.

Techniques: Generated

Journal: Frontiers in Neuroscience

Article Title: A Deep Network Model on Dynamic Functional Connectivity With Applications to Gender Classification and Intelligence Prediction

doi: 10.3389/fnins.2020.00881

Figure Lengend Snippet: Prediction performance of fluid intelligence and crystallized intelligence. (A) The correlations between predicted and observed intelligence scores for the CNN + LSTM and support vector machine (SVM) models. Note that the CNN + LSTM model exhibits the highest correlation scores for both tasks. Each subject is represented by one dot, and 95% confidence interval for the best-fit line is represented by the gray area which is used to assess the predictive power of the model. (B) Comparison between mean MAE across 10 testing folds for the CNN + LSTM and the SVM model. Lower is better. Bars refer to the mean accuracy of all testing folds, and error bars refer to their standard error. Note that the CNN + LSTM model is statistically better than the SVM model (* p < 0.05).

Article Snippet: For comparing with the CNN + LSTM model, we used linear SVM and linear epsilon SVR models (LIBSVM toolbox in Matlab ) based on dynamic characteristic of dFC (dFC-Str, which refers to the overall strength of dFC) , to achieve gender classification and intelligence prediction with the same 10-fold cross-validation strategies.

Techniques: Plasmid Preparation, Comparison

Journal: Frontiers in Neuroscience

Article Title: A Deep Network Model on Dynamic Functional Connectivity With Applications to Gender Classification and Intelligence Prediction

doi: 10.3389/fnins.2020.00881

Figure Lengend Snippet: Model performance of rs-fMRI based gender classification and intelligence prediction tasks in some recent studies.

Article Snippet: For comparing with the CNN + LSTM model, we used linear SVM and linear epsilon SVR models (LIBSVM toolbox in Matlab ) based on dynamic characteristic of dFC (dFC-Str, which refers to the overall strength of dFC) , to achieve gender classification and intelligence prediction with the same 10-fold cross-validation strategies.

Techniques: Surround Optical-fiber Immunoassay