A deep multi-task learning approach to identifying mummy berry infection sites, the disease stage, and severity.

Introduction: Mummy berry is a serious disease that may result in up to 70 percent of yield loss for lowbush blueberries. Practical mummy berry disease detection, stage classification and severity estimation remain great challenges for computer vision-based approaches because images taken in lowbush blueberry fields are usually a mixture of different plant parts (leaves, bud, flowers and fruits) with a very complex background. Specifically, typical problems hindering this effort included data scarcity due to high manual labelling cost, tiny and low contrast disease features interfered and occluded by healthy plant parts, and over-complicated deep neural networks which made deployment of a predictive system difficult. Methods: Using real and raw blueberry field images, this research proposed a deep multi-task learning (MTL) approach to simultaneously accomplish three disease detection tasks: identification of infection sites, classification of disease stage, and severity estimation. By further incorporating novel superimposed attention mechanism modules and grouped convolutions to the deep neural network, enabled disease feature extraction from both channel and spatial perspectives, achieving better detection performance in open and complex environments, while having lower computational cost and faster convergence rate. Results: Experimental results demonstrated that our approach achieved higher detection efficiency compared with the state-of-the-art deep learning models in terms of detection accuracy, while having three main advantages: 1) field images mixed with various types of lowbush blueberry plant organs under a complex background can be used for disease detection; 2) parameter sharing among different tasks greatly reduced the size of training samples and saved 60% training time than when the three tasks (data preparation, model development and exploration) were trained separately; and 3) only one-sixth of the network parameter size (23.98M vs. 138.36M) and one-fifteenth of the computational cost (1.13G vs. 15.48G FLOPs) were used when compared with the most popular Convolutional Neural Network VGG16. Discussion: These features make our solution very promising for future mobile deployment such as a drone carried task unit for real-time field surveillance. As an automatic approach to fast disease diagnosis, it can be a useful technical tool to provide growers real time disease information that can prevent further disease transmission and more severe effects on yield due to fruit mummification.

Guaranteed cost control via dynamic output feedback for networked interval type-2 T-S fuzzy system with hybrid communication mechanism.

This paper aims to investigate the guaranteed cost control via dynamic output feedback for nonlinear networked control systems (NCSs) with consideration of hybrid communication mechanism, data dropout and bounded disturbance. Interval type-2 (IT2) Takagi-Sugeno (T-S) fuzzy model is utilized to describe the nonlinear system with parameter uncertainties. To enhance bandwidth utilization and improve control performance, a hybrid communication mechanism involving both event-triggered mechanism (ETM) and time-triggered mechanism (TTM) is proposed. Two Bernoulli processes are invoked to describe the switching between two triggering mechanisms, and the data dropout phenomenon in communication network, respectively. The quadratic boundedness (QB) technique is employed to specify the closed-loop stability of a bounded disturbance networked system. The sufficient conditions for the stability of the system and the presence of a dynamic output feedback guaranteed cost controller are presented. In addition, the problem of controller design is converted to a convex optimization problem that can be tackled by linear matrix inequalities (LMIs) technique. At last, simulation experiment is carried out to explicate the availability and usefulness of the designed controller.

Batesian mimicry or general food deception? An evolutionary game between plants for pollinator services.

Floral food deception is a well-known phenomenon which is not thoroughly understood. Particularly, it is unclear what drives a plant towards Batesian mimicry or towards generalized food deception. We analysed the evolutionary game between a Model species with nectar-secreting flowers and a Deceiver species that provides no nectar who share pollinators for reproduction. We focused our analysis on the effect of similarity of floral signals between participating plants and on costs of nectar production. We defined payoffs in the game between Models and Deceivers as the stationary visitation frequencies to participating species with different signal similarities and nectar costs. Therefore, fitness payoff of each strategy was a product of complex interactions between plant species composing the community and the pollinators visiting them. Our model provides a unified framework in which consequences of Model species interaction with different deception modes can be compared. Our findings suggest that plant-pollinator systems, like other mutualistic systems, are prone to exploitation, and that exploitation may persist at a large range of conditions. We showed that floral similarity, and thus, pollinators' ability to discriminate between Model and deceptive species, governs the stability of Batesian mimicry, while pollinator switching and sampling behaviour enables the persistence of general food deception.

Optimization of drug scheduling for cancer chemotherapy with considering reducing cumulative drug toxicity.

An improved optimal drug scheduling model with considering two control drugs is proposed and the Gauss pseudospectral-based optimization method is studied to decrease the tumor size and drug toxicity in this work. Firstly, the Dexrazoxane drug, which has significant clinical effect to reduce the toxicity of the anticancer drug, is introduced. By analyzing the growth kinetics model of cancer chemotherapy, the toxicity reduction drug is regarded as the second input in the cancer dynamic equations. Correspondingly, the drug scheduling optimization problem with particular optimization goal and necessary constraints is established. Next, a model transformation technique is proposed to reduce the complexity of dynamic equations. With deriving the Gaussian time grid discretization detailly, the Gauss pseudospectral method (GPM)-based cancer chemotherapy drug scheduling algorithm is presented to test the performance of the proposed model within different rates. Finally, the implementation structure of drug scheduling optimization is given in detail. To test and validate the performance of proposed chemotherapy model, extensive simulation results and comparative evaluation are carried out on a specific mathematical model. Simulation results show that the improved optimization model is superior to other literature studies, resulting in the average improvement of performance index by 66.54% and revealing the significant guiding property for cancer chemotherapy.

Anomaly detection for blueberry data using sparse autoencoder-support vector machine.

High-dimensional space includes many subspaces so that anomalies can be hidden in any of them, which leads to obvious difficulties in abnormality detection. Currently, most existing anomaly detection methods tend to measure distances between data points. Unfortunately, the distance between data points becomes more similar as the dimensionality of the input data increases, resulting in difficulties in differentiation between data points. As such, the high dimensionality of input data brings an obvious challenge for anomaly detection. To address this issue, this article proposes a hybrid method of combining a sparse autoencoder with a support vector machine. The principle is that by first using the proposed sparse autoencoder, the low-dimensional features of the input dataset can be captured, so as to reduce its dimensionality. Then, the support vector machine separates abnormal features from normal features in the captured low-dimensional feature space. To improve the precision of separation, a novel kernel is derived based on the Mercer theorem. Meanwhile, to prevent normal points from being mistakenly classified, the upper limit of the number of abnormal points is estimated by the Chebyshev theorem. Experiments on both the synthetic datasets and the UCI datasets show that the proposed method outperforms the state-of-the-art detection methods in the ability of anomaly detection. We find that the newly designed kernel can explore different sub-regions, which is able to better separate anomaly instances from the normal ones. Moreover, our results suggested that anomaly detection models suffer less negative effects from the complexity of data distribution in the space reconstructed by those layered features than in the original space.

A molecular dynamics study of phospholipid membrane electroporation induced by bipolar pulses with different intervals.

The mechanism of changes in cell electroporation (EP) during the intervals of bipolar pulses is still unclear, and few studies have investigated the effect of the intervals at the molecular level. In this study, EP induced by bipolar pulses (BP) with different intervals was investigated using all-atom molecular dynamics simulations. Firstly, EP was formed during the positive pulses of 2 ns and 0.5 V nm-1, then the effects of various intervals of 0, 1, 5, and 10 ns on EP evolution were investigated, and the dynamic changes of different degrees of EP induced by the following negative pulses of 2 ns and 0.5 V nm-1 were analyzed. The elimination effect of intervals was determined and it was related to the degrees of EP and the time of intervals. At the last moment of the intervals the phospholipid membrane was classified and quantitatively defined in three states according to the degrees of EP, namely, Resealing, Destabilizing and Retaining states. These states appeared due to the combined effect of both the positive pulse and the interval, and the states represent the degrees of EP which had different responses after applying the negative pulse. These results can improve our understanding of the fundamental mechanism of BP-induced EP.

Effects of climate change on vegetation dynamics of the Qinghai-Tibet Plateau, a causality analysis using empirical dynamic modeling.

Given the vital role of the Qinghai-Tibet Plateau (QTP) as water tower in Asia and regulator for regional and even global climate, the relationship between climate change and vegetation dynamics on it has received considerable focused attention. Climate change may influence the vegetation growth on the plateau, but clear empirical evidence of such causal linkages is sparse. Herein, using datasets CRU-TS v4.04 and AVHHR NDVI from 1981 to 2019, we quantify causal effects of climate factors on vegetation dynamics with an empirical dynamical model (EDM) -- a nonlinear dynamical systems analysis approach based on state-space reconstruction rather than correlation. Results showed the following: (1) climate change promotes the growth of vegetation on the QTP, and specifically, this favorable influence of temperature is stronger than precipitation's; (2) the direction and strength of climate effects on vegetation varied over time, and the effects are seasonally different; (3) a significant increase in temperature and a slight increase in precipitation are beneficial to vegetation growth, specifically, NDVI will increase within 2% in the next 40 years with the climate trend of warming and humidity. Besides the above results, another interesting finding is that the two seasons in which precipitation strongly influence vegetation in the Three-River Source region (part of the QTP) are spring and winter. This study provides insights into the mechanisms by which climate change affects vegetation growth on the QTP, aiding in the modeling of vegetation dynamics in future scenarios.

An Unsupervised Feature Extraction Approach Based on Self-Expression.

Feature extraction algorithms lack good interpretability during the projection learning. To solve this problem, an unsupervised feature extraction algorithm, that is, block diagonal projection (BDP), based on self-expression is proposed. Specifically, if the original data are projected into a low-dimensional subspace by a feature extraction algorithm, although the data may be more compact, the new features obtained may not be as explanatory as the original sample features. Therefore, by imposing L2,1 norm constraint on the projection matrix, the projection matrix can be of row sparsity. On one hand, discriminative features can be selected to make the projection matrix to be more interpretable. On the other hand, irrelevant or redundant features can be suppressed. The proposed model integrates feature extraction and selection into one framework. In addition, since self-expression can well excavate the correlation between samples or sample features, the unsupervised feature extraction task can be better guided using this property between them. At the same time, the block diagonal representation regular term is introduced to directly pursue the block diagonal representation. Thus, the accuracy of pattern recognition tasks such as clustering and classification can be improved. Finally, the effectiveness of BDP in linear dimensionality reduction and classification is proved on various reference datasets. The experimental results show that this algorithm is superior to previous feature extraction counterparts.

All-atom molecular dynamics simulations of the combined effects of different phospholipids and cholesterol content on electroporation.

The electroporation mechanism could be related to the composition of the plasma membrane, and the combined effect of different phospholipid molecules and cholesterol content on electroporation has rarely been studied nor conclusions drawn. In this paper, we applied all-atom molecular dynamics (MD) simulations to study the effects of phospholipids and cholesterol content on bilayer membrane electroporation. The palmitoyloleoylphosphatidylcholine (POPC) model, palmitoyloleoylphosphatidylethanolamine (POPE) model, and a 1 : 1 mixed model of POPC and POPE called PEPC, were the three basic models used. An electric field of 0.45 V nm-1 was applied to nine models, which were the three basic models, each with three different cholesterol content values of 0%, 24%, and 40%. The interfacial water molecules moved under the electric field and, once the first water bridge formed, the rest of the water molecules would dramatically flood into the membrane. The simulation showed that a rapid rise in the Z-component of the average dipole moment of the interfacial water molecules (Z-DM) indicated the occurrence of electroporation, and the same increment of Z-DM represented a similar change in the size of the water bridge. With the same cholesterol content, the formation of the first water bridge was the most rapid in the POPC model, regarding the average electroporation time (t ep), and the average t ep of the PEPC model was close to that of the POPE model. We speculate that the differences in membrane thickness and initial number of hydrogen bonds of the interfacial water molecules affect the average t ep for different membrane compositions. Our results reveal the influence of membrane composition on the electroporation mechanism at the molecular level.

A novel autoencoder approach to feature extraction with linear separability for high-dimensional data.

Feature extraction often needs to rely on sufficient information of the input data, however, the distribution of the data upon a high-dimensional space is too sparse to provide sufficient information for feature extraction. Furthermore, high dimensionality of the data also creates trouble for the searching of those features scattered in subspaces. As such, it is a tricky task for feature extraction from the data upon a high-dimensional space. To address this issue, this article proposes a novel autoencoder method using Mahalanobis distance metric of rescaling transformation. The key idea of the method is that by implementing Mahalanobis distance metric of rescaling transformation, the difference between the reconstructed distribution and the original distribution can be reduced, so as to improve the ability of feature extraction to the autoencoder. Results show that the proposed approach wins the state-of-the-art methods in terms of both the accuracy of feature extraction and the linear separabilities of the extracted features. We indicate that distance metric-based methods are more suitable for extracting those features with linear separabilities from high-dimensional data than feature selection-based methods. In a high-dimensional space, evaluating feature similarity is relatively easier than evaluating feature importance, so that distance metric methods by evaluating feature similarity gain advantages over feature selection methods by assessing feature importance for feature extraction, while evaluating feature importance is more computationally efficient than evaluating feature similarity.

Improved predictive control approach to networked control systems based on quantization dependent Lyapunov function.

This paper considers model predictive control (MPC) for the linear discrete-time systems in the presence of packet loss, quantization and actuator saturation. Compared with the previous work ([45]), this paper presents an improved networked MPC approach for networked control systems (NCSs) by applying the quantization dependent Lyapunov function (QDLF) method which leads to less conservative results. The additional improvement is made by placing the heavier weighting on the system corresponding to the actual linear feedback law and choosing the relative weighting on the actual and auxiliary feedback laws which further improves the control performance over the existing method. It is shown that the closed-loop stability is guaranteed and a quantized state-feedback controller is derived by solving the infinite horizon optimization problem. Moreover, this method is further extended to multiple-input case. A numerical example is given to illustrate the effectiveness of the proposed approach.

Hesperidin ameliorates behavioral impairments and neuropathology of transgenic APP/PS1 mice.

In addition to cognitive impairments, deficits in non-cognitive behaviors are also common neurological sequelae in Alzheimer's disease and its animal models. Hesperidin, a flavanone glycoside found abundantly in citrus fruits, was orally given (100 mg/kg body weight) to 5-month-old transgenic APP/PS1 mice, a mouse model of cerebral amyloidosis for Alzheimer's disease. After a relatively short-term treatment of 10 days, hesperidin significantly restored deficits in non-cognitive nesting ability and social interaction. Further immunohistochemical analysis showed significantly attenuated ß-amyloid deposition, plaque associated APP expression, microglial activation and TGF-ß immunoreactivity in brains of APP/PS1 mice, which suggests that ameliorated behavioral impairments might be attributable to reduced Aß deposition and attenuated neuro-inflammatory reaction. Additionally, efficient anti-inflammatory effects of hesperidin were confirmed in vitro. Our findings suggest that hesperidin might be a potential candidate for the treatment of AD or even other neurodegenerative diseases.

Constrained off-line synthesis approach of model predictive control for networked control systems with network-induced delays.

This paper investigates the off-line synthesis approach of model predictive control (MPC) for a class of networked control systems (NCSs) with network-induced delays. A new augmented model which can be readily applied to time-varying control law, is proposed to describe the NCS where bounded deterministic network-induced delays may occur in both sensor to controller (S-A) and controller to actuator (C-A) links. Based on this augmented model, a sufficient condition of the closed-loop stability is derived by applying the Lyapunov method. The off-line synthesis approach of model predictive control is addressed using the stability results of the system, which explicitly considers the satisfaction of input and state constraints. Numerical example is given to illustrate the effectiveness of the proposed method.