• Publications
  • Influence
Complex Embeddings for Simple Link Prediction
TLDR
This work makes use of complex valued embeddings to solve the link prediction problem through latent factorization, and uses the Hermitian dot product, the complex counterpart of the standard dot product between real vectors. Expand
Constructing Datasets for Multi-hop Reading Comprehension Across Documents
TLDR
A novel task to encourage the development of models for text understanding across multiple documents and to investigate the limits of existing methods, in which a model learns to seek and combine evidence — effectively performing multihop, alias multi-step, inference. Expand
Knowledge Graph Completion via Complex Tensor Factorization
TLDR
The approach based on complex embeddings is arguably simple, as it only involves a Hermitian dot product, the complex counterpart of the standard dot product between real vectors, whereas other methods resort to more and more complicated composition functions to increase their expressiveness. Expand
Frustratingly Short Attention Spans in Neural Language Modeling
TLDR
This paper proposes a neural language model with a key-value attention mechanism that outputs separate representations for the key and value of a differentiable memory, as well as for encoding the next-word distribution that outperforms existing memory-augmented neural language models on two corpora. Expand
UCL Machine Reading Group: Four Factor Framework For Fact Finding (HexaF)
TLDR
This system is a four stage model consisting of document retrieval, sentence retrieval, natural language inference and aggregation that achieved a FEVER score of 62.52% on the provisional test set (without additional human evaluation), and 65.41%" on the development set. Expand
Neural Random Forests
TLDR
This work reformulates the random forest method of Breiman (2001) into a neural network setting, and proposes two new hybrid procedures that are called neural random forests, which both predictors exploit prior knowledge of regression trees for their architecture. Expand
Achieving Verified Robustness to Symbol Substitutions via Interval Bound Propagation
TLDR
This work studies text classification under synonym replacements or character flip perturbations and modifies the conventional log-likelihood training objective to train models that can be efficiently verified, which would otherwise come with exponential search complexity. Expand
Reducing Sentiment Bias in Language Models via Counterfactual Evaluation
TLDR
This paper quantifies sentiment bias by adopting individual and group fairness metrics from the fair machine learning literature, and proposes embedding and sentiment prediction-derived regularization on the language model’s latent representations. Expand
Casting Random Forests as Artificial Neural Networks (and Profiting from It)
TLDR
Formalizing a connection between Random Forests and ANN allows exploiting the former to initialize the latter, and parameter optimization within the ANN framework yields models that are intermediate betweenRF and ANN, and achieve performance better than RF and ANN on the majority of the UCI datasets used for benchmarking. Expand
Crowdsourcing Multiple Choice Science Questions
TLDR
This work presents a novel method for obtaining high-quality, domain-targeted multiple choice questions from crowd workers by leveraging a large corpus of domain-specific text and a small set of existing questions and shows that humans cannot distinguish the crowdsourced questions from original questions. Expand
...
1
2
...