Search for: All records

Real-world applications often involve irregular time series, for which the time intervals between successive observations are non-uniform. Irregularity across multiple features in a multi-variate time series further results in a different subset of features at any given time (i.e., asynchronicity). Existing pre-training schemes for time-series, however, often assume regularity of time series and make no special treatment of irregularity. We argue that such irregularity offers insight about domain property of the data—for example, frequency of hospital visits may signal patient health condition—that can guide representation learning. In this work, we propose PrimeNet to learn a self-supervised representation for irregular multivariate time-series. Specifically, we design a timesensitive contrastive learning and data reconstruction task to pre-train a model. Irregular time-series exhibits considerable variations in sampling density over time. Hence, our triplet generation strategy follows the density of the original data points, preserving its native irregularity. Moreover, the sampling density variation over time makes data reconstruction difficult for different regions. Therefore, we design a data masking technique that always masks a constant time duration to accommodate reconstruction for regions of different sampling density. We learn with these tasks using unlabeled data to build a pre-trained model and fine-tune on a downstream task with limited labeled data, in contrast with existing fully supervised approach for irregular time-series, requiring large amounts of labeled data. Experiment results show that PrimeNet significantly outperforms state-of-the-art methods on naturally irregular and asynchronous data from Healthcare and IoT applications for several downstream tasks, including classification, interpolation, and regression. 

Various hardware accelerators have been developed for energy-efficient and real-time inference of neural networks on edge devices. However, most training is done on high-performance GPUs or servers, and the huge memory and computing costs prevent training neural networks on edge devices. This paper proposes a novel tensor-based training framework, which offers orders-of-magnitude memory reduction in the training process. We propose a novel rank-adaptive tensorized neural network model, and design a hardware-friendly low-precision algorithm to train this model. We present an FPGA accelerator to demonstrate the benefits of this training method on edge devices. Our preliminary FPGA implementation achieves 59× speedup and 123× energy reduction compared to embedded CPU, and 292× memory reduction over a standard full-size training. 

Due to environmental concerns and the increasing drive towards miniaturization of electronic circuits and devices, lead-free ferroelectric films with low leakage current and robust ferroelectric and piezoelectric properties are highly desired. The preferred alternative, BaTiO 3 , is non-toxic and has ferroelectric properties, but its high leakage current, poor ferroelectricity and piezoelectricity and low Curie temperature of ∼130 °C in thin film form are obstacles for high-temperature practical applications. Here, we report that a negative-pressure-driven enhancement of ferroelectric Curie temperature and effective piezoelectric coefficient are achieved in (111)-oriented BaTiO 3 nanocomposite films. The enhanced ferroelectric and piezoelectric properties in the emergent monoclinic BaTiO 3 are attributed to the sharp vertical interface and 3D tensile strain that develops upon interspersing stiff and self-assembled vertical Sm 2 O 3 nanopillars through the film thickness. Our work also demonstrates that fabricating oxide films through (111)-oriented epitaxy opens up new avenues for the creation of new phase components and exploration of novel functionalities for developing oxide quantum electronic devices.