Athiya Deviyani

I am a second-year Masters student at Carnegie Mellon University pursuing an MS in Artificial Intelligence and Innovation. Last summer, I completed an applied machine learning research internship in the Search Quality team at Apple, where I implemented a GraphSAGE Graph Neural Network to suggest related questions for Siri search and web answers.

In the summer of 2021, I graduated top of my class with a high first from The University of Edinburgh with a BSc (Hons) in Artificial Intelligence and Computer Science. My undergraduate honors thesis on mitigating dataset bias in computer vision by artificially generating data from minority classes using GANs under the supervision of Ajitha Rajan received an outstanding paper classification. My research interests lie in machine learning and computational linguistics with a focus in fairness, accountability, transparency, and ethics (FATE).

In spring 2022, I conducted research under Alan W. Black to develop tools and explore methodologies for generalizable text normalization for speech systems. Previously, I have worked as a spring or summer intern at Goldman Sachs, Google, YouTube, Amazon, and JP Morgan. I’ve also conducted research on computational social science and taught some machine learning classes in my undergraduate university.

news

May 23, 2022	Our paper “Text Normalization for Speech Systems for All Languages” is accepted to the Speech for Social Good Workshop at INTERSPEECH 2022 🎉
May 23, 2022	Started my AI/ML internship with the Information Intelligence (Search Quality) team at Apple 👩🏻‍💻🍎
Sep 26, 2021	Won BigRed//Hacks at Cornell University with my new friends at CMU 🤓
Aug 30, 2021	Started my Masters in Artificial Intelligence and Innovation at Carnegie Mellon University 🇺🇸
Jul 2, 2021	Graduated top of my class with a first class honors from The University of Edinburgh with a BSc in Artificial Intelligence and Computer Science! 👩🏻‍🎓🏴󠁧󠁢󠁳󠁣󠁴󠁿
Jun 22, 2021	My undergraduate thesis “Assessing Dataset Bias in Computer Vision” received an outstanding classification (blind double marking) from the School of Informatics at the University of Edinburgh 🎉

selected publications

INTERSPEECH

Text Normalization for Speech Systems for All Languages

Deviyani, Athiya, and Black, Alan W.

In INTERSPEECH’22 Workshop on Speech for Social Good Jun 2022

Abs PDF

Most text-to-speech systems suffer from the limitation that its inputs should be a set of strings of characters with standard pronunciation, and struggle when given input is in the form of symbols, numbers, or abbreviations that often occur in real text. One of the most common ways to address this problem is to automatically map non-standard words to standard words using statistical, neural and rule-driven methods. However, despite the significant efforts of normalizing such words, there is just too much variability in existing corpora such that it is extremely challenging to capture edge cases. In this work, we propose a tool which aids data collection from (non-programmer) native speakers to allow numbers and other common non-standard words to be mapped to standard words that can be pronounced correctly by a synthesizer, while addressing related problems such as identifying common non-standard words appear in text and how do we ask questions from native speakers to get sufficient information to allow a useful normalization of non-standard words.
UoE

Assessing Dataset Bias in Computer Vision

Deviyani, Athiya

In The University of Edinburgh Outstanding Undergraduate Dissertations Jun 2021

Abs PDF

A biased dataset is a dataset that generally has attributes with an uneven class distribution. These biases have the tendency to propagate to the models that train on them, often leading to a poor performance in the minority class. In this project, we will explore the extent to which various data augmentation methods alleviate intrinsic biases within the dataset. We will apply several augmentation techniques on a sample of the UTKFace dataset, such as undersampling, geometric transformations, variational autoencoders (VAEs), and generative adversarial networks (GANs). We then trained a classifier for each of the augmented datasets and evaluated their performance on the native test set and on external facial recognition datasets. We have also compared their performance to the state-of-the-art attribute classifier trained on the FairFace dataset. Through experimentation, we were able to find that training the model on StarGAN-generated images led to the best overall performance. We also found that training on geometrically transformed images lead to a similar performance with a much quicker training time. Additionally, the best performing models also exhibit a uniform performance across the classes within each attribute. This signifies that the model was also able to mitigate the biases present in the baseline model that was trained on the original training set. Finally, we were able to show that our model has a better overall performance and consistency on age and ethnicity classification on multiple datasets when compared with the FairFace model.