No abstract

No abstract

Neural-networks based image restoration methods tend to use low-resolution image patches for training. Although higher-resolution image patches can provide more global information, state-of-the-art methods cannot utilize them due to their huge GPU memory usage, as well as the instable training process. However, plenty of studies have shown that global information is crucial for image restoration tasks like image demosaicing and enhancing. In this work, we propose a HighEr-Resolution Network (HERN) to fully learning global information in high-resolution image patches. To achieve this, the HERN employs two parallel paths to learn image features in two different resolution, respectively. By combining global-aware features and multi-scale features, our HERN is able to learn global information with feasible GPU memory usage. Besides, we introduce a progressive training method to solve the instability issue and accelerate model convergence. On the task of image demosaicing and enhancing, our HERN achieves state-of-the-art performance on the AIM2019 RAW to RGB mapping challenge. The source code of our implementation is available at https://github.com/MKFMIKU/ RAW2RGBNet.

Conventional lane change methods directly collected steering angle data via onboard sensors to accurately capture the actions of individual drivers. We can hardly use such methods to collect massive data from examinees, because of time and financial costs. In order to retrieve common steering behaviors for lots of drivers, we propose a method to retrieve common Discretionary Lane Change (DLC) steering characteristics from trajectory data. The key technique of this new method is solving an inverse problem that converts the measured trajectory into the unmeasured steering maneuvers under the assumed vehicle movement dynamics. We find that most normal DLC trajectories in the Next Generation Simulation (NGSIM) datasets could be well reproduced by a simple target heading angle preview control model. This finding sheds important light into driver behavior study and better explains how human control vehicles. Based on these findings, we can non-intrusively evaluate driving performance or physiological states of drivers based on online roadside monitoring data (e.g. the data collected from roadside video cameras). This opens a promising field of applications for enhancing driving safety.

Recent explorations of large-scale pre-trained language models (PLMs) have revealed the power of PLMs with huge amounts of parameters, setting off a wave of training everlarger PLMs. However, it requires tremendous computational resources to train a largescale PLM, which may be practically unaffordable. In addition, existing large-scale PLMs are mainly trained from scratch individually, ignoring that many well-trained PLMs are available. To this end, we explore the question how could existing PLMs benefit training large-scale PLMs in future. Specifically, we introduce a pre-training framework named "knowledge inheritance" (KI) and explore how could knowledge distillation serve as auxiliary supervision during pre-training to efficiently learn larger PLMs. Experimental results demonstrate the superiority of KI in training efficiency. We also conduct empirical analyses to explore the effects of teacher PLMs' pre-training settings, including model architecture, pre-training data, etc. Finally, we show that KI could be applied to domain adaptation and knowledge transfer. The implementation is publicly available at https://github.com/thunlp/ Knowledge-Inheritance.