• Publications
  • Influence
Don’t Parse, Generate! A Sequence to Sequence Architecture for Task-Oriented Semantic Parsing
TLDR
A unified architecture based on Sequence to Sequence models and Pointer Generator Network to handle both simple and complex queries is proposed and achieves state of the art performance on three publicly available datasets. Expand
iPlug: Decentralised dispatch of distributed generation
TLDR
A decentralised controller iPlug is designed, which involves voltage-sensitive back-off of the level of solar capacity injection into the grid and is inspired by the CSMA protocol from Networking literature, and could in principle be incorporated into either the inverter controls or battery and home energy management control systems. Expand
Unsupervised Parsing with S-DIORA: Single Tree Encoding for Deep Inside-Outside Recursive Autoencoders
TLDR
S-DIORA, an improved variant of DIORA that encodes a single tree rather than a softly-weighted mixture of trees by employing a hard argmax operation and a beam at each cell in the chart, is introduced. Expand
Exploring Transfer Learning For End-to-End Spoken Language Understanding
TLDR
This work proposes an E2E system that is designed to jointly train on multiple speech-to-text tasks, such as ASR (speech-transcription) and SLU ( speech-hypothesis), and text- to- text tasks,such as NLU (text- Hypothesis) and calls it the Audio-Text All-Task (AT-AT) Model, and shows that it beats the performance of E1E models trained on individual tasks. Expand
Improved Pretraining for Domain-specific Contextual Embedding Models
TLDR
This work investigates methods to mitigate catastrophic forgetting during domain-specific pretraining of contextual embedding models such as BERT, DistilBERT, and RoBERTa and proposes the use of two continual learning techniques (rehearsal and elastic weight consolidation) to improve domain- specific training. Expand
Estimating return on investment for grid scale storage within the economic dispatch framework
TLDR
This paper forms an optimization problem to trade off the economic advantage of using storage devices versus its loss of life, and uses detailed formulations to estimate loss of battery life to estimate capital expenditure (capex) on battery storage, on a per-day basis. Expand
A context vector regression based approach for demand forecasting in district heating networks
TLDR
A data analytics based modeling framework, which uses a context vector based approach for forecasting energy consumption, is proposed, which is effective in identifying the appropriate context combination that can help explain historical consumption as observed for a given consumer. Expand
Learning Latent Space Representations to Predict Patient Outcomes: Model Development and Validation
TLDR
The results support the applicability of CLOUT for real-world clinical use in identifying patients at high risk of mortality. Expand
From multiple views to single view: a neural network approach
TLDR
This work proposes an approach to multi-view learning based on a recently proposed autoencoder model called Predictive AutoEncoder (PAE) and shows that the proposed approach performs better than the existing MVL approaches like co-training and Canonical Correlation Analysis. Expand
Compressing Transformer-Based Semantic Parsing Models using Compositional Code Embeddings
TLDR
This work proposes to learn compositional code embeddings to greatly reduce the sizes of BERT-base and RoBERTa-base, and applies the technique to DistilBERT, ALBERT- base, and ALberT-large, three already compressed BERT variants which attain similar state-of-the-art performances on semantic parsing with much smaller model sizes. Expand
...
1
2
...