Research laboratory on the mechanics of smart materials and structures, Zhejiang University

Abstract

Large-scale pre-training has brought unimodal fields such as computer vision and natural language processing to a new era. Following this trend, the size of multimodal learning models constantly increases, leading to an urgent need to reduce the massive computational cost of finetuning these models for downstream tasks. In this paper, we propose an efficient and flexible multimodal fusion method, namely PMF, tailored for fusing unimodally pretrained transformers. Specifically, we first present a modular multimodal fusion framework that exhibits high flexibility and facilitates mutual interactions among different modalities. In addition, we disentangle vanilla prompts into three types in order to learn different optimizing objectives for multimodal learning. It is also worth noting that we propose to add prompt vectors only on the deep layers of the unimodal transformers, thus significantly reducing the training memory usage. Experiment results show that our proposed method achieves comparable performance to several other multimodal finetuning methods with less than 3% trainable parameters and up to 66% saving of training memory usage.

Soldered joints of piezoelectric products

“…We can clearly see the through defects in the glass-metal soldered joint after opening the products. The control is carried out under a Shuttlepix P-400R digital microscope in ultraviolet light (LI et al, 2019;YANG et al, 2019;WANG et al, 2018;CHENG et al, 2018).…”

Science and education in the form 4.0: public policy and organization based on human and artificial intellectual capital

“…Ardito et al (2019) note the top-priority role of universities in knowledge management in the “smart” city projects – and universities are to be “smart” as well. Li et al (2019) offer a model of “smart” research laboratory for mechanics of intellectual materials and constructions by the example of Zhejiang University. Fortes et al (2019) offer a vision of campus as a “smart city” (by the example of University of Malaga).…”

Design and optimization of an Electromagnetic- Piezoelectric Composite Vibration Generator

“…In the environment of mechanical motion, the form of motion such as vibration is very common, and the vibration energy collected in the environment is often enough to drive some electronic equipment to work. Using the piezoelectric effect of functional material [1], a piezoelectric integrated structure is designed to convert mechanical energy into electrical energy. By converting and collecting the generated electrical energy, it can provide power for various low-power electronic systems.…”

“…(1) y (t)--Vibration displacement of the external environment foundation; z(t)--Relative vibration displacement of the mass block; ξ--Damping ratio, the system efficient elastic coefficient, m = ∆m +0.236 M[7], ∆m is the mass of mass blocks, M is the cantilever beam mass.…”