Joint Transformer/RNN Architecture for Gesture Typing in Indic Languages

Abstract

Gesture typing is a method of typing words on a touch-based keyboard by creating a continuous trace passing through the relevant keys. This work is aimed at developing a keyboard that supports gesture typing in Indic languages. We begin by noting that when dealing with Indic languages, one needs to cater to two different sets of users: (i) users who prefer to type in the native Indic script (Devanagari, Bengali, etc.) and (ii) users who prefer to type in the English script but want the transliterated output in the native script. In both cases, we need a model that takes a trace as input and maps it to the intended word. To enable the development of these models, we create and release two datasets. First, we create a dataset containing keyboard traces for 193,658 words from 7 Indic languages. Second, we curate 104,412 English-Indic transliteration pairs from Wikidata across these languages. Using these datasets we build a model that performs path decoding, transliteration and transliteration correction. Unlike prior approaches, our proposed model does not make co-character independence assumptions during decoding. The overall accuracy of our model across the 7 languages varies from 70-95%.

Input-specific Attention Subnetworks for Adversarial Detection

“…This is because character-level changes make the token invalid, i.e, the model treats it as a unknown token absent in the vocabulary. Since this changes the input embedding sequence more dramatically (Biju et al, 2020)…”

Input-specific Attention Subnetworks for Adversarial Detection

“…This is because character-level changes make the token invalid, i.e, the model treats it as a unknown token absent in the vocabulary. Since this changes the input embedding sequence more dramatically (Biju et al, 2020), small deviations from standard gating patterns are sufficient to mislead the model leading to fewer refereeing heads. Since introducing synonym and embedding based perturbations change the embeddings input to the model by a smaller extent, larger deviations from the gating pattern are required to block or pass selective chunks of information to mislead the model.…”