A test for the absence of aliasing or white noise in two-dimensional locally stationary wavelet processes

Article Computer Science, Information Systems

Computing Systems for Autonomous Driving: State of the Art and Challenges

Liangkai Liu et al.

Summary: The key to the success of current autonomous driving systems lies in making reliable real-time decisions, while the real traffic environment is too complex for these systems to fully grasp. This article introduces state-of-the-art computing systems for autonomous driving and highlights the challenges to be addressed, aiming to draw attention from the industry and inspire further research.

IEEE INTERNET OF THINGS JOURNAL (2021)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

Impact of Aliasing on Generalization in Deep Convolutional Networks

Cristina Vasconcelos et al.

Summary: This research investigates the impact of aliasing on generalization in Deep Convolutional Networks and proposes the insertion of non-trainable low-pass filters at key locations to mitigate aliasing, leading to substantial improvements in generalization on i.i.d. and out-of-distribution conditions. State-of-the-art results are achieved on ImageNet-C and Meta-Dataset without introducing additional trainable parameters, using default hyper-parameters of open source codebases.

2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021) (2021)

Add to Collection

Proceedings Paper Acoustics

HOW CONVOLUTIONAL NEURAL NETWORKS DEAL WITH ALIASING

Antonio H. Ribeiro et al.

Summary: In this paper, the ability of convolutional neural networks to counteract aliasing is explored through two examples, highlighting the importance of redundancies in intermediate channels for successful task completion, and indicating that while CNNs are capable of implementing anti-aliasing filters in theory, they do not completely prevent aliasing in intermediate layers.

2021 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP 2021) (2021)

Add to Collection

Article Computer Science, Artificial Intelligence

WaveCNet: Wavelet Integrated CNNs to Suppress Aliasing Effect for Noise-Robust Image Classification

Qiufu Li et al.

Summary: By integrating discrete wavelet transform (DWT) with convolutional neural networks (CNN), the proposed approach enhances the noise robustness and adversarial robustness, achieving higher accuracy in image classification tasks. Decomposing feature maps using DWT into low-frequency and high-frequency components, the model generates robust high-level features and eliminates data noises, leading to improved performance on ImageNet dataset and adversarial attacks.

IEEE TRANSACTIONS ON IMAGE PROCESSING (2021)

Add to Collection

Article Engineering, Electrical & Electronic