Search

Recent empirical evidence indicates that bond excess returns can be predicted using machine learning models. While the predictive power of machine learning models is intriguing, they typically lack transparency. We introduce SHapley Additive exPlanations (SHAP), a state-of-the-art explainable artificial technique, to open the black box of these models. Our analysis identifies the key determinants that drive the predictions of bond excess returns in machine learning models and how these determinants are related to bond excess returns. Thereby, our approach facilitates an in-depth interpretation of the predictions of bond excess returns made by machine learning models. [ABSTRACT FROM AUTHOR]

This study presents a novel approach to fabricate highly-structured 3D vertically-aligned carbon nanotubes (VA-CNTs) architectures, and further integrates 3D VA-CNTs with MEMS surface micromachining to fabricate suspended 3D CNTs structures. The merits of presented approach are as follows: (1) highly-structured 3D VA-CNTs stacks can be achieved by predefining multilayered Al/Fe catalyst films; (2) only single run of the pyrolysis CVD process is required to grow multilayered 3D VA-CNTs stacks; (3) the design flexibility of the microstructure in out-of-plane direction can be increased by the variation layers of CNTs stacks. Preliminary fabrication results implement various 3D VA-CNTs architectures. Furthermore, 3D CNTs stacks have been integrated with MEMS surface micromachining to demonstrate different thickness of 3D CNTs suspended cantilevers and probe. The out-of-plane resonant frequency has also been calibrated to characterize different stiffness of the 3D CNTs suspended devices. [ABSTRACT FROM PUBLISHER]

This paper addresses the feasibility of the so-called Modal Signaling (MS) transmission scheme from a stochastic viewpoint. MS has been proposed for crosstalk mitigation over interconnects and is based on the encoding of signals onto fundamental transmission-line modes. However, the design of drivers and receivers strongly depends on the physical characteristics of the channel. In this paper, the impact of random variations of these properties on MS effectiveness is efficiently analyzed by means of Polynomial Chaos (PC) technique. [ABSTRACT FROM PUBLISHER]

In this paper, the feasibility of Modal Signaling (MS) transmission scheme in presence of dielectric variability is addressed from a stochastic standpoint. Recently, MS has been proposed to reduce crosstalk over high-speed interconnects. Being it based on the encoding of signals onto fundamental transmission-line modes, the correct operation requires an accurate knowledge of the physical interconnect parameters. The impact of random variations in the dielectric permittivity on MS effectiveness is analyzed by means of an efficient tool for the stochastic simulation of interconnects, namely Polynomial Chaos (PC) technique. [ABSTRACT FROM PUBLISHER]

The reliability is an important issue for IC devices. Since moisture, mobile ions, ultraviolet, and other hostile environment will affect the device performance and lifetime. In our study, a robust conformal superhydrophobic (SH) coating will be applied to improve the reliability of the device. SH epoxy/nano silica composites were spin-coated on triple track resistors (TTR). An O2 plasma treatment was used to generate surface roughness on the cured epoxy. SH surfaces were produced after functionalization with perfluorooctyl trichlorosilane of low surface energy. The effect of plasma etching time on superhydrophobicty was investigated. A water contact angle of 170° and contact angle hysteresis of 2° were achieved when the etching time increased to 15 min. The leakage currents under bias for SH coated TTR were measured with aging time. The reliability test showed stable leakage current for 1000 hours, and the self-cleaning performance of the SH surface successfully protects the device. [ABSTRACT FROM PUBLISHER]