• Publications
  • Influence
Pathologies of Neural Models Make Interpretations Difficult
TLDR
This work uses input reduction, which iteratively removes the least important word from the input, to expose pathological behaviors of neural models: the remaining words appear nonsensical to humans and are not the ones determined as important by interpretation methods.
The galE Gene of Campylobacter jejuni Is Involved in Lipopolysaccharide Synthesis and Virulence
TLDR
The LPS analysis of wild-type, galE, and complementedgalE Salmonella strains showed that the C. jejuni galE gene could restore the smooth wild- typeSalmonella LPS.
Universal Adversarial Triggers for NLP
Adversarial examples highlight model vulnerabilities and are useful for evaluation and interpretation. We define universal adversarial triggers: input-agnostic sequences of tokens that trigger a
Customizing Triggers with Concealed Data Poisoning
TLDR
This work develops a new data poisoning attack that allows an adversary to control model predictions whenever a desired trigger phrase is present in the input.
What can AI do for me?: evaluating machine learning interpretations in cooperative play
TLDR
This work designs a grounded, realistic human-computer cooperative setting using a question answering task, Quizbowl, and proposes an evaluation of interpretation on a real task with real human users, where the effectiveness of interpretation is measured by how much it improves human performance.
Knowledge-Based Semantic Embedding for Machine Translation
TLDR
This paper builds and formulate a semantic space to connect the source and target languages, and applies it to the sequence-to-sequence framework to propose a Knowledge-Based Semantic Embedding (KBSE) method.
Implicit Distortion and Fertility Models for Attention-based Encoder-Decoder NMT Model
TLDR
This work proposes new variations of attention-based encoder-decoder and compares them with other models on machine translation to resolve problems caused by the lack of distortion and fertility models.
Improving Attention Modeling with Implicit Distortion and Fertility for Machine Translation
TLDR
Two novel models to improve attention-based neural machine translation are proposed, a recurrent attention mechanism as a implicit distortion model, and a fertility conditioned decoder as an implicit fertility model that significantly improve both the alignment and translation quality.
Concealed Data Poisoning Attacks on NLP Models
TLDR
This work develops a new data poisoning attack that allows an adversary to control model predictions whenever a desired trigger phrase is present in the input.
ERNIE 3.0 Titan: Exploring Larger-scale Knowledge Enhanced Pre-training for Language Understanding and Generation
TLDR
A hundred-billion-parameter model called ERNIE 3.0 Titan with up to 260 billion parameters is trained, which is the largest Chinese dense pre-trained model so far and outperforms the state-of-the-art models on 68 NLP datasets.
...
1
2
3
4
5
...