84 citations · 161 across the 14 of their papers we have counts for
14 papers
Spatial Entropy as an Inductive Bias for Vision Transformers
Elia Peruzzo, Enver Sangineto, Yahui Liu +4
Recent work on Vision Transformers (VTs) showed that introducing a local inductive bias in the VT architecture helps reducing the number of samples necessary for training. However,…
Masked Jigsaw Puzzle: A Versatile Position Embedding for Vision Transformers
Bin Ren, Yahui Liu, Yue Song +4
Position Embeddings (PEs), an arguably indispensable component in Vision Transformers (ViTs), have been shown to improve the performance of ViTs on many vision tasks. However, PEs…
Efficient Training of Visual Transformers with Small Datasets
Yahui Liu, Enver Sangineto, Wei Bi +3
Visual Transformers (VTs) are emerging as an architectural paradigm alternative to Convolutional networks (CNNs). Differently from CNNs, VTs can capture global relations between im…
A Batch Normalized Inference Network Keeps the KL Vanishing Away
Qile Zhu, Jianlin Su, Wei Bi +4
Variational Autoencoder (VAE) is widely used as a generative model to approximate a model's posterior on latent variables by combining the amortized variational inference and deep…
Response-Anticipated Memory for On-Demand Knowledge Integration in Response Generation
Zhiliang Tian, Wei Bi, Dongkyu Lee +4
Neural conversation models are known to generate appropriate but non-informative responses in general. A scenario where informativeness can be significantly enhanced is Conversing…
Learning to Select Bi-Aspect Information for Document-Scale Text Content Manipulation
Xiaocheng Feng, Yawei Sun, Bing Qin +5
In this paper, we focus on a new practical task, document-scale text content manipulation, which is the opposite of text style transfer and aims to preserve text styles while alter…
Improving Knowledge-aware Dialogue Generation via Knowledge Base Question Answering
Jian Wang, Junhao Liu, Wei Bi +4
Neural network models usually suffer from the challenge of incorporating commonsense knowledge into the open-domain dialogue systems. In this paper, we propose a novel knowledge-aw…
Relevance-Promoting Language Model for Short-Text Conversation
Xin Li, Piji Li, Wei Bi +2
Despite the effectiveness of sequence-to-sequence framework on the task of Short-Text Conversation (STC), the issue of under-exploitation of training data (i.e., the supervision si…
Learning to Customize Model Structures for Few-shot Dialogue Generation Tasks
Yiping Song, Zequn Liu, Wei Bi +2
Training the generative models with minimal corpus is one of the critical challenges for building open-domain dialogue systems. Existing methods tend to use the meta-learning frame…
-differential agreement: A Parallel Data Sorting Mechanism for Distributed Information Processing System
Wei Bi, Xiangyu Liu, Maolin Zheng
The order of the input information plays a very important role in a distributed information processing system (DIPS). This paper proposes a novel data sorting mechanism named the ε…
Generating Multiple Diverse Responses for Short-Text Conversation
Jun Gao, Wei Bi, Xiaojiang Liu +2
Neural generative models have become popular and achieved promising performance on short-text conversation tasks. They are generally trained to build a 1-to-1 mapping from the inpu…
An Accelerated Method for Message Propagation in Blockchain Networks
Wei Bi, Huawei Yang, Maolin Zheng
Blockchain is based on a P2P network, supporting decentralized consensus of current cryptocurrencies. Since bitcoin and altcoins all utilize an underlying blockchain, they are ther…
Language Style Transfer from Sentences with Arbitrary Unknown Styles
Yanpeng Zhao, Wei Bi, Deng Cai +3
Language style transfer is the problem of migrating the content of a source sentence to a target style. In many of its applications, parallel training data are not available and so…
A Secure Multiple Elliptic Curves Digital Signature Algorithm for Blockchain
Wei Bi, Xiaoyun Jia, Maolin Zheng
Most cryptocurrency systems or systems based on blockchain technology are currently using the elliptic curves digital signature algorithm (ECDSA) on the secp256k1 curve, which is s…