Search for: All records

We consider transferability estimation, the prob- lem of estimating how well deep learning models transfer from a source to a target task. We focus on regression tasks, which received little previous attention, and propose two simple and computa- tionally efficient approaches that estimate trans- ferability based on the negative regularized mean squared error of a linear regression model. We prove novel theoretical results connecting our ap- proaches to the actual transferability of the optimal target models obtained from the transfer learning process. Despite their simplicity, our approaches significantly outperform existing state-of-the-art regression transferability estimators in both accu- racy and efficiency. On two large-scale keypoint re- gression benchmarks, our approaches yield 12% to 36% better results on average while being at least 27% faster than previous state-of-the-art methods. 

Triboluminescence (TL) is a phenomenon of light emission induced by impact, stress, fracture, or an applied mechanical force. This phenomenon can be used to detect, evaluate, and predict mechanical failures in composites. In this report, we utilized manganese-doped zinc-sulphide (ZnS: Mn) and Polystyrene (PS) composite to fabricate a TL functional part via additive manufacturing. The morphology of the particles inside the polymer matrix were studied using scanning electron microscopy and micro CT scan. Thermoanalytical techniques such as differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA) were carried out to evaluate the thermal transitions and degradation of the composites. The mechanoluminescence performance of the printed samples is evaluated by three-point flexural test and observed to depend on processing conditions that can be utilized to achieve a strong light signal at different mechanical loads. The polymer composite fabrication and processing reduced particle size, enhanced particle dispersion, and altered the mechanical properties of the polymer to help increase the mechanoluminescence response up to 10 times in the 3D printed parts. The unique mechanoluminescence properties of 3D printed luminescent composite have great potential for structural monitoring applications. 

The demand for additively manufactured polymer composites with increased specific properties and functional microstructure has drastically increased over the past decade. The ability to manufacture complex designs that can maximize strength while reducing weight in an automated fashion has made 3D-printed composites a popular research target in the field of engineering. However, a significant amount of understanding and basic research is still necessary to decode the fundamental process mechanisms of combining enhanced functionality and additively manufactured composites. In this review, external field-assisted additive manufacturing techniques for polymer composites are discussed with respect to (1) self-assembly into complex microstructures, (2) control of fiber orientation for improved interlayer mechanical properties, and (3) incorporation of multi-functionalities such as electrical conductivity, self-healing, sensing, and other functional capabilities. A comparison between reinforcement shapes and the type of external field used to achieve mechanical property improvements in printed composites is addressed. Research has shown the use of such materials in the production of parts exhibiting high strength-to-weight ratio for use in aerospace and automotive fields, sensors for monitoring stress and conducting electricity, and the production of flexible batteries.