---

Optical interconnects as appropriate alternatives for electrical interconnects in the computer chips and boards can be realized by CMOS-based integrated silicon photonics. Dielectric slot waveguide, as one of the newest optical waveguide structures, can form the infrastructure of the passive and active components in these integrated circuits. The passive components have the linear behavior. In order to realize the all-optical active components such as laser, amplifier, and modulator we can use the nonlinear effects in the silicon photonics waveguides. On the other hand, Si-nc:SiO2 as a new material, has a stronger nonlinear property than Si. The results of the full-wave analyzes of the slot waveguide in the linear and nonlinear regimes show that the slot region of this waveguide can be filled with the Si-nc:SiO2 and also realize a high optical intensity. Therefore, this waveguide intensifies the nonlinear behaviors by two factors.

---

 In this paper an observer-based robust fault estimation scheme is proposed for a special class of Lipchitz nonlinear systems where the disturbances and faults are assumed to be coupled with the main system states. In the considered model of system, fault is assumed to enter both of the state and output equations as an unmeasured nonlinear function and coupled with the states. The disturbances and the uncertainties are considered as nonlinear functions coupled with the states. To the best of the authors’ knowledge these conditions have not been previously considered in related papers. In the proposed approach, a Luenberger observer is designed for the estimation of faults and states of system simultaneously. The effect of system disturbances is attenuated with the L2  norm. The necessary conditions for the existence of such observer is expressed in the form of Linear Matrix Inequality. The Lipchitz constant of the nonlinear function is obtained by solving the proposed Linear Matrix Inequality. Finally, the performance of the proposed method is simulated on a three-phase induction motor. The results indicate good performance of the proposed method.
 

---

Insufficient training data is one of the main challenges of utilizing deep Convolutional Neural Networks (CNNs) for Environmental Sound Classification (ESC). As a promising solution, Transfer Learning (TL) has addressed this issue by adapting a network pre-trained on a large-scale dataset to the target task. In this paper, we demonstrate that not all neurons/kernels of every layer in CNN networks are equally utilized to process the inputs of different classes, but there is a specific subgroups of neurons/kernels in every layer that play the key role in classification of every output class. Based on this observation and due to similarities that exist between feature spaces of some source and target classes, we propose to concentrate the fine-tuning process only on those neurons/kernels that do need changes and have the greatest impact on misclassifying target data. To identify these neurons/kernels, we pose a nested optimization problem for which we propose an effective evolutionary approach as solution.  Compared to the conventional fine-tuning approach, our proposed method achieves absolute improvements of about 1.9% and 2.3% in accuracy on ESC-50 and DCASE-17, respectively; remarkable improvements produced not by adding augmented data but with a more efficient utilization of knowledge stored in the pre-trained network. It is noteworthy that the computation time overhead of the proposed evolutionary method is rather small (about one third of the time required to train the model from scratch.