• Publications
  • Influence
GPT-GNN: Generative Pre-Training of Graph Neural Networks
TLDR
GPT-GNN introduces a self-supervised attributed graph generation task to pre-train a GNN so that it can capture the structural and semantic properties of the graph. Expand
  • 23
  • 6
  • PDF
A Transformer-based Approach for Source Code Summarization
TLDR
We explore the Transformer model that uses a self-attention mechanism and has shown to be effective in capturing long-range dependencies. Expand
  • 18
  • 5
  • PDF
Robustness Verification for Transformers
TLDR
Robustness verification that aims to formally certify the prediction behavior of neural networks has become an important tool for understanding the behavior of a given model. Expand
  • 18
  • 2
  • PDF
Automatic Perturbation Analysis for Scalable Certified Robustness and Beyond
TLDR
In this paper, we develop an automatic framework to enable perturbation analysis on any neural network structures, by generalizing existing LiRPA algorithms such as CROWN to operate on general computational graphs. Expand
  • 9
  • 2
  • PDF
Provable, Scalable and Automatic Perturbation Analysis on General Computational Graphs
TLDR
We develop an automatic framework to enable perturbation analysis on any neural network structures, by generalizing exiting LiRPA algorithms such as CROWN to operate on general computational graphs. Expand
  • 2
  • 2
Retrofitting Contextualized Word Embeddings with Paraphrases
TLDR
We propose a simple and effective paraphrase-aware retrofitting (PAR) method that is applicable to arbitrary pretrained contextualized embeddings. Expand
  • 9
  • 1
  • PDF
Automatic Perturbation Analysis on General Computational Graphs
TLDR
Linear relaxation based perturbation analysis for neural networks, which aims to compute tight linear bounds of output neurons given a certain amount of input perturbations, has become a core component in robustness verification and certified defense. Expand
  • 4
  • 1
TRAP: A Three-Way Handshake Server for TCP Connection Establishment
Distributed denial of service attacks have become more and more frequent nowadays. In 2013, a massive distributed denial of service (DDoS) attack was launched against Spamhaus causing the service toExpand
  • 7
  • 1
Efficient Contextual Representation Learning With Continuous Outputs
TLDR
We propose a new approach for learning contextual representation models with continuous outputs that achieves a 4-fold speedup and achieves competitive performance on downstream tasks. Expand
  • 3
  • 1
  • PDF
SentiBERT: A Transferable Transformer-Based Architecture for Compositional Sentiment Semantics
TLDR
We propose SentiBERT, a variant of BERT that effectively captures compositional sentiment semantics. Expand
  • 14
  • PDF
...
1
2
3
4
5
...