In this paper, a novel model for encrypting text messages using time varying delayed Hopfield neural network and a posterior DNA cryptographic model is proposed. The chaotic neural network applied here is used to generate a binary sequence which is later passed to a permutation function and generate the key for the first level encryption. The plaintext is converted to a corresponding binary sequence after a conversion to ASCI value and encrypted by switching of chaotic neural network maps and a permutation function which is dependent on the binary sequence generated from the chaotic neural network. An additional DNA cryptographic model is used over the cipher text obtained from the first level encryption to robust the security of the proposed model.

Traveling Salesman Problem (TSP) is the most well-known combinatorial optimization real-world problem. TSP is also very popular to check proficiency in any newly developed optimization method. In addition, the optimization methods, which are developed for other tasks (e. g., numerical optimization), also test their proficiency in TSP. This study investigates a new technique to solve TSP based on a recently developed optimization technique stimulated through the foraging conduct of spider monkeys. Standard Spider Monkey Optimization (SMO) is established for numerical optimization which has six phases and each one has a different purpose. In this study, SMO is modified and updated to solve TSP; and Swap Operators (SOs), and Swap Sequence (SS) are considered to adapt SMO for TSP. In the proposed method, each spider monkey is considered as a TSP solution and SS is considered to update the solution. SS is an arrangement of several SOs in which each one holds two particular positions of a solution that might be swapped to make a new solution. All SOs of a SS is applied on a specific tour maintaining order and thus ramifications of the SS change the TSP tour into another one. The SOs are generated using the experience of a specific spider monkey as well as the experience of other members (local leader, global leader, or randomly selected spider monkey) of the group. The proposed strategy has been examined on a huge number of benchmark TSPs and final consequences are compared to other prominent methods. SMO shows 11 best result out of 15 benchmark TSP problem compare to Ant Colony Optimization (ACO) and Velocity Tentative PSO (VTPSO). Experimental consequences show that the proposed strategy is a decent technique to solve TSP.

The recognition of pathological voice is considered a difficult task for speech analysis. Moreover, otolaryngologists needed to rely on oral communication with patients to discover traces of voice pathologies like dysphonia that are caused by voice alteration of vocal folds and their accuracy is between 60%–70%. To enhance detection accuracy and reduce processing speed of dysphonia detection, a novel approach is proposed in this paper. We have leveraged Linear Discriminant Analysis (LDA) to train multiple Machine Learning (ML) models for dysphonia detection. Several ML models are utilized like Support Vector Machine (SVM), Logistic Regression, and K-nearest neighbor (K-NN) to predict the voice pathologies based on features like Mel-Frequency Cepstral Coefficients (MFCC), Fundamental Frequency (F0), Shimmer (%), Jitter (%), and Harmonic to Noise Ratio (HNR). The experiments were performed using Saarbrucken Voice Database (SVD) and a privately collected dataset. The K-fold cross-validation approach was incorporated to increase the robustness and stability of the ML models. According to the experimental results, our proposed approach has a 70% increase in processing speed over Principal Component Analysis (PCA) and performs remarkably well with a recognition accuracy of 95.24% on the SVD dataset surpassing the previous best accuracy of 82.37%. In the case of the private dataset, our proposed method achieved an accuracy rate of 93.37%. It can be an effective non-invasive method to detect dysphonia.

Machine learning (ML) suffers from a persistent and critical flaw: adversarial examples. Many new forms of adversarial example attacks have been invented and many narrow defenses have been proposed. Unfortunately, no defensive approach can withstand current attacks. We hypothesize that ML model robustness can be improved with approaches that delineate the data-point-sparse latent space between data-dense regions of a model’s classification space as a barrier class. We introduce one such defense, PadNet, that builds a barrier class using a combination of training samples that mix multiple classes together. It leverages this barrier class to separate decision boundaries between benign classes with regions of padding. PadNet then implements a gradient regularization strategy that penalizes gradients in the direction of the barrier class, causing the decision boundary to draw tighter around training samples increasing boundary thickness between classes. We evaluate PadNet against a sampling of the most effective state-of-the-art attacks, demonstrating that it offers significant robustness and reliability compared to current defenses. We also test it against adaptive attacks and find that PadNet remains robust against them.