Compared to many other dense prediction tasks, e.g., semantic segmentation, it is the arbitrary number of instances that has made instance segmentation much more challenging. In order to predict a …
Modeling event patterns is a central task in a wide range of disciplines. In applications such as studying human activity patterns, events often arrive clustered with sporadic and long periods of i…
Point cloud instance segmentation has achieved huge progress with the emergence of deep learning. However, these methods are usually data-hungry with expensive and time-consuming dense point cloud …
Weakly supervised temporal action localization aims at learning the instance-level action pattern from the video-level labels, where a significant challenge is action-context confusion. To overcome…
In this work, we report on a high-power all-fiber ultrafast laser system at 2.0 μm that delivers femtosecond pulses with a fundamental repetition rate of 3.0 GHz and a maximum output power of 33.1…
Transfer learning (TL) has been demonstrated its feasibility on fast remodeling for fiber nonlinearity equalization. It will be very efficient with fine-tuning rather than retraining when the chann…
Forward error correction (FEC) performance down to 1e-15 bit error rate (BER) of a open FEC code (OFEC), which was recently proposed for the 800G inter-data center interconnect (DCI) standard, is v…
By employing the mode decomposition method to study the relative mode evolution characteristics during the stimulated Raman scattering (SRS) process, a numerical model of the core-pumped Raman effe…
We propose and demonstrate a novel concept of Brillouin fiber swept laser with narrow linewidth based on mode following effect. Utilizing the synergy of the Brillouin gain spectrum and cavity mode,…
In this article, we develop uniform inference methods for the conditional mode based on quantile regression. Specifically, we propose to estimate the conditional mode by minimizing the derivative o…