Haytham Fayek

AI/ML Research/Science | Senior Lecturer @RMIT | Alumnus @Meta @Facebook @Oculus @Worley_ecr @RMIT @UTPOfficial | Views own.

[Haytham Fayek @HaythamFayek]

I am looking for a PhD student to work on an exciting NLP/LLM project. Must be an Australian citizen and be able to start in 2025. Full scholarship and stipend available. Reach out if you’re interested or for more details.

[Nannan (Amber) Huang @ ACL 2026 @AmberNNHuang]

🎉 Excited to share that our paper "Less Is More? Examining Fairness in Pruned Large Language Models for Summarising Opinions" has been accepted at @emnlpmeeting #EMNLP2025

👉 https://t.co/YJJwJAxqD0

@HaythamFayek

[Jay Yang @Jayyanginspires]

this is the way

[Haytham Fayek @HaythamFayek]

@random_walker I have been thinking and struggling with this over the past few years. Everyone here seems to be on the same boat. I would love to know how some people solved this.

[Haytham Fayek @HaythamFayek]

We will present our spotlight work, SWAP-NAS: Sample-Wise Activation Patterns for Ultra-fast NAS (https://t.co/vG37D6Vtp3), on Wed 8 May 10:45am-12:45pm (https://t.co/UlOYs9P1Yq).

[Haytham Fayek @HaythamFayek]

At @iclr_conf in Vienna this week to present SWAP-NAS: Sample-Wise Activation Patterns for Ultra-fast NAS (https://t.co/vG37D6Vtp3). This is my first large non-virtual conference in a while, so I am excited to catch up with many friends and a week of interesting conversations.

[Haytham Fayek @HaythamFayek]

@LakeBrenden Congratulations Brenden! Well-deserved.

[Haytham Fayek @HaythamFayek]

We establish a methodology to quantify bias in abstractive opinion summarisation models and study political bias in pretrained LLMs and various fine-tuning methods. We find models express intrinsic bias, and fine-tuning these models to summarise social media text amplified bias.

[Haytham Fayek @HaythamFayek]

Our paper, Bias in Opinion Summarisation from Pre-training to Adaptation: A Case Study in Political Bias, led by @AmberNNHuang , was accepted at @eaclmeeting #EACL2024.

👉 https://t.co/ukfURlty13

[Haytham Fayek @HaythamFayek]

Our paper, SWAP-NAS: Sample-Wise Activation Patterns for Ultra-fast NAS, led by Yameng Peng, was accepted at @iclr_conf 2024 (spotlight).

We propose a training-free metric for Neural Architecture Search (NAS) leading to SOTA NAS on ImageNet in 9 mins.

https://t.co/OcwgZx4arA

[Haytham Fayek @HaythamFayek]

@dr1337 Or updating the nvidia driver.

[Haytham Fayek @HaythamFayek]

What a time to be alive! Snoop Dog talks about AI Risk and Geoff Hinton!

https://t.co/v9WcZvYRXs

[Haytham Fayek @HaythamFayek]

Jarry (https://t.co/EMKK7fXe3a) will present the SCARY dataset as part of a panel discussion on datasets for causal learning at @CLeaR_2022'23, April 12, 4-5pm CET (https://t.co/9gNPYrQEdI).

[Haytham Fayek @HaythamFayek]

Be Afraid 👻! We're releasing a new synthetic benchmark and dataset for causal discovery, The Structurally Complex with Additive Parent Causality (SCARY) Dataset, led by Jarry Chen, accepted at @CLeaR_2022'23.

Paper: https://t.co/7TmSAypbGg
Data: https://t.co/7woerMTXaS

More 👇

[Haytham Fayek @HaythamFayek]

The SCARY dataset is composed of 240 subsets with 40 unique configurations that emphasise practical issues of unfaithfulness, causal sufficiency, and selection bias, with various densities and causal mechanisms, ideal for benchmarking causal discovery algorithms.

[Haytham Fayek @HaythamFayek]

Our paper, A Case for Personalised Non-Player Character Companion Design, led by @emmajanepretty, was accepted at IJHCI, puts forward a framework for personalised non-player character (NPC) companions using psychophysiological and performance data.

👉 https://t.co/bCT667OCWt

[Haytham Fayek @HaythamFayek]

Our work, led by @emmajanepretty, puts forward a case and a framework for personalised non-player character (NPC) companions using psychophysiological and performance data, is out.

Title: A Case for Personalised Non-Player Character Companion Design
👉 https://t.co/lAlEyvQ7z0

[Haytham Fayek @HaythamFayek]

Authors: @renghp, @emmajanepretty, Fabio Zambetta, & Ron van Schyndel.

[Haytham Fayek @HaythamFayek]

I am pleased that our work on "Evoking empathy with visually impaired people through an augmented reality embodiment experience", led by Renan Guarese, was accepted at @IEEEVR. In Memory of our colleague (@RMITComputing) and co-author Ron van Schyndel who sadly passed recently.👇

[Haytham Fayek @HaythamFayek]

We propose a multi-sensory interactive experience that allows sighted users to embody having a visual impairment whilst using assistive technologies. We show that the experience increases empathy and sympathy towards the Blind & Visually Impaired (BVI) community.