Deep Model Compression and Inference Speedup of Sum-Product Networks on Tensor Trains

Sum-product networks (SPNs) constitute an emerging class of neural networks with clear probabilistic semantics and superior inference speed over other graphical models. This brief reveals an important connection between SPNs and tensor trains (TTs), leading to a new canonical form which we call tensor SPNs (tSPNs). Specifically, we...

Overdriving reliability of chip scale packaged LEDs: Quantitatively analyzing the impact of component

The objective of this study is to quantitatively evaluate the impacts of LED components on the overdriving reliability of high power white LED chip scale packages (CSPs). The reliability tests under room temperature are conducted over 1000 h in this study on CSP LEDs with overdriving currents. A novel method is proposed to investigate the...

Thermal inductance in GaN devices

Using the analogue of the electric inductance, we reveal the properties of the thermal inductance in GaN-based light-emitting diode devices by testing their transient thermal behaviors. We find that the devices exhibit a transient thermal response under step-down or step-up currents and observe notable inductive phenomena of the temperature...

Molecular modeling of temperature dependence of solubility parameters for amorphous polymers

A molecular modeling strategy is proposed to describe the temperature (T) dependence of solubility parameter (?) for the amorphous polymers which exhibit glass-rubber transition behavior. The commercial forcefield “COMPASS” is used to support the atomistic simulations of the polymer. The temperature dependence behavior of ? for the polymer is...

Evaluation and selection of sensing materials for carbon dioxide (CO2) sensor by molecular modeling

We report a molecular modeling study to evaluate and select conducting polymers (CPs) as the sensing materials of carbon dioxide (CO2) sensor. The interaction between polymer and gas and the adsorption of the gas molecules in the polymer matrix are investigated. Polymers considered for this work include emeraldine base polyaniline (EB–PANI) and...

Forcefields based molecular modeling on the mechanical and physical properties of emeraldine base polyaniline

Molecular dynamics (MD) and molecular mechanical (MM) analysis are carried out to provide reliable and accurate model for emeraldine base polyaniline. This study validate the forcefields and model with the physical and mechanical properties of the polyaniline. The temperature effects on non-bond energy, potential energy and solubility parameter...