Searcharxiv⌕ Search

arXiv subjects

Matei Ionita

Publications and source records attributed to Matei Ionita.

2 recordsLinked to original sources

Spectral Networks and Non-abelianization

We generalize the non-abelianization of Gaiotto-Moore-Neitzke from the case of $SL(n)$ and $GL(n)$ to arbitrary reductive algebraic groups. This gives a map between a moduli space of certain $N$-shifted weakly $W$-equivariant $T$-local systems on an open subset of a cameral cover $\tilde{X}\rightarrow X$ to the moduli space of $G$-local systems on a punctured Riemann surface $X$. For classical groups, we give interpretations of these moduli spaces using spectral covers. Non-abelianization uses a set of lines on the Riemann surface $X$ called a spectral network, defined using a point in the Hitchin base. We show that these lines are related to trajectories of quadratic differentials on quotients of $\tilde{X}$. We use this to describe some of the generic behaviour of lines in a spectral network.

math.AG↗

Resolving Gendered Ambiguous Pronouns with BERT

Pronoun resolution is part of coreference resolution, the task of pairing an expression to its referring entity. This is an important task for natural language understanding and a necessary component of machine translation systems, chat bots and assistants. Neural machine learning systems perform far from ideally in this task, reaching as low as 73% F1 scores on modern benchmark datasets. Moreover, they tend to perform better for masculine pronouns than for feminine ones. Thus, the problem is both challenging and important for NLP researchers and practitioners. In this project, we describe our BERT-based approach to solving the problem of gender-balanced pronoun resolution. We are able to reach 92% F1 score and a much lower gender bias on the benchmark dataset shared by Google AI Language team.

cs.CL↗