Graph Neural Networks with Learnable Structural and Positional Representations

vijaydwivedi75/gnn-lspe 15 Oct 2021

An approach to tackle this issue is to introduce Positional Encoding (PE) of nodes, and inject it into the input layer, like in Transformers.

Knowledge Graphs Recommendation Systems

55
1.09 stars / hour

Taming Visually Guided Sound Generation

v-iashin/SpecVQGAN 17 Oct 2021

In this work, we propose a single model capable of generating visually relevant, high-fidelity sounds prompted with a set of frames from open-domain videos in less time than it takes to play it on a single GPU.

Audio Generation

64
0.92 stars / hour

TLDR: Twin Learning for Dimensionality Reduction

naver/tldr 18 Oct 2021

In this paper, we unify these two families of approaches from the angle of manifold learning and propose TLDR, a dimensionality reduction method for generic input spaces that is porting the simple self-supervised learning framework of Barlow Twins to a setting where it is hard or impossible to define an appropriate set of distortions by hand.

Dimensionality Reduction Representation Learning +1

31
0.83 stars / hour

NeuralBlox: Real-Time Neural Representation Fusion for Robust Volumetric Mapping

ethz-asl/neuralblox 18 Oct 2021

We present a novel 3D mapping method leveraging the recent progress in neural implicit representation for 3D reconstruction.

3D Reconstruction

24
0.67 stars / hour

Mengzi: Towards Lightweight yet Ingenious Pre-trained Models for Chinese

langboat/mengzi 13 Oct 2021

Although pre-trained models (PLMs) have achieved remarkable improvements in a wide range of NLP tasks, they are expensive in terms of time and resources.

77
0.57 stars / hour

FlexConv: Continuous Kernel Convolutions with Differentiable Kernel Sizes

rjbruin/flexconv 15 Oct 2021

In this work, we propose FlexConv, a novel convolutional operation with which high bandwidth convolutional kernels of learnable kernel size can be learned at a fixed parameter cost.

Sequential Image Classification Time Series

28
0.55 stars / hour

P-Tuning v2: Prompt Tuning Can Be Comparable to Fine-tuning Universally Across Scales and Tasks

thudm/p-tuning-v2 14 Oct 2021

Prompt tuning, which only tunes continuous prompts with a frozen language model, substantially reduces per-task storage and memory usage at training.

Language Modelling

65
0.52 stars / hour

Mix3D: Out-of-Context Data Augmentation for 3D Scenes

kumuji/mix3d 5 Oct 2021

Since scene context helps reasoning about object semantics, current works focus on models with large capacity and receptive fields that can fully capture the global context of an input 3D scene.

3D Semantic Segmentation

73
0.47 stars / hour

CIPS-3D: A 3D-Aware Generator of GANs Based on Conditionally-Independent Pixel Synthesis

PeterouZh/CIPS-3D 19 Oct 2021

The style-based GAN (StyleGAN) architecture achieved state-of-the-art results for generating high-quality images, but it lacks explicit and precise control over camera poses.

Image Generation Transfer Learning

65
0.46 stars / hour

CoAtNet: Marrying Convolution and Attention for All Data Sizes

xmu-xiaoma666/External-Attention-pytorch 9 Jun 2021

Transformers have attracted increasing interests in computer vision, but they still fall behind state-of-the-art convolutional networks.

 Ranked #1 on Image Classification on ImageNet (using extra training data)

Image Classification

2,147
0.46 stars / hour