Daewon Lee

Minseok Kim, Woosung Choi, Jaehwa Chung et al.

Recently, many methods based on deep learning have been proposed for music source separation. Some state-of-the-art methods have shown that stacking many layers with many skip connections improve the SDR performance. Although such a deep and complex architecture shows outstanding performance, it usually requires numerous computing resources and time for training and evaluation. This paper proposes a two-stream neural network for music demixing, called KUIELab-MDX-Net, which shows a good balance of performance and required resources. The proposed model has a time-frequency branch and a time-domain branch, where each branch separates stems, respectively. It blends results from two streams to generate the final estimation. KUIELab-MDX-Net took second place on leaderboard A and third place on leaderboard B in the Music Demixing Challenge at ISMIR 2021. This paper also summarizes experimental results on another benchmark, MUSDB18. Our source code is available online.

Daewon Lee

Steganography is one of the information hiding techniques, which conceals secret messages in cover media. Digital image and audio are the most studied cover media for steganography. However, so far, there is no research on steganography to utilize complex networks as cover media. To investigate the possibility and feasibility of complex networks as cover media for steganography, we introduce steganography of complex networks through three algorithms: BIND, BYMOND, and BYNIS. BIND hides two bits of a secret message in an edge, while BYMOND encodes a byte in an edge, without changing the original network structures. Encoding simulation experiments for the networks of Open Graph Benchmark demonstrated BIND and BYMOND can successfully hide random messages in the edge lists. BYNIS synthesizes edges by generating node identifiers from a given message. The degree distribution of stego network synthesized by BYNIS was mostly close to a power-law. Steganography of complex networks is expected to have applications such as watermarking to protect proprietary datasets, or sensitive information hiding for privacy preservation.

Woosung Choi, Minseok Kim, Jaehwa Chung et al.

Singing Voice Separation (SVS) tries to separate singing voice from a given mixed musical signal. Recently, many U-Net-based models have been proposed for the SVS task, but there were no existing works that evaluate and compare various types of intermediate blocks that can be used in the U-Net architecture. In this paper, we introduce a variety of intermediate spectrogram transformation blocks. We implement U-nets based on these blocks and train them on complex-valued spectrograms to consider both magnitude and phase. These networks are then compared on the SDR metric. When using a particular block composed of convolutional and fully-connected layers, it achieves state-of-the-art SDR on the MUSDB singing voice separation task by a large margin of 0.9 dB. Our code and models are available online.

Selim Engin, Eric Mitchell, Daewon Lee et al.

We consider the problem of planning views for a robot to acquire images of an object for visual inspection and reconstruction. In contrast to offline methods which require a 3D model of the object as input or online methods which rely on only local measurements, our method uses a neural network which encodes shape information for a large number of objects. We build on recent deep learning methods capable of generating a complete 3D reconstruction of an object from a single image. Specifically, in this work, we extend a recent method which uses Higher Order Functions (HOF) to represent the shape of the object. We present a new generalization of this method to incorporate multiple images as input and establish a connection between visibility and reconstruction quality. This relationship forms the foundation of our view planning method where we compute viewpoints to visually cover the output of the multi-view HOF network with as few images as possible. Experiments indicate that our method provides a good compromise between online and offline methods: Similar to online methods, our method does not require the true object model as input. In terms of number of views, it is much more efficient. In most cases, its performance is comparable to the optimal offline case even on object classes the network has not been trained on.