Xiaojun Chang

Professor in Artificial Intelligence at School of Computer Science, University of Technology Sydney

[Xiaojun Chang @cxj273]

Our lab established the first public CT scan report dataset, so as to promote the development of AI technology for CT report generation to screen and detect COVID-19 patients.

More information about this dataset can be found at https://t.co/VWJZrvBrEM.

[Xiaojun Chang @cxj273]

@muttler Well deserved!!!

[Flora Salim @flosalim]

The large scale FFA-IR dataset with >10K reports and >1M images with English and Chinese annotations enables training of XAI, medical report generation and contributes to medical image understanding.
The link to the dataset: https://t.co/VCbWK9PcsS

@cxj273 @karinv
#NeurIPS21

[Flora Salim @flosalim]

Thrilled to share our accepted #NeurIPS21 @NeurIPSConf paper “FFA-IR: Towards an Explainable and Reliable Medical Report Generation Benchmark, with @cxj273 @karinv. https://t.co/Ff611xSAnM
FFA-IR is the largest medical imaging dataset with bilingual annotation.

1/2🧵
#XAI #ML

[PyTorch @PyTorch]

This tutorial will introduce compute and data-efficient transformers and provide a step-by-step to create your own Vision Transformers. Through this guide, you'll be able to train state of the art results for classification in both computer vision & NLP.

https://t.co/d3bc7ijeBJ

[Xiaojun Chang @cxj273]

Arxiv: https://t.co/MvZ7cXsxA9

Code: https://t.co/qSVGGKXRnG

[Xiaojun Chang @cxj273]

I do not regularly advertise our work, but I really want to share our recent work named BossNAS. We have released the code of this work. We made the following contributions in this work:

[Xiaojun Chang @cxj273]

(2) we present HyTra search space, a fabric-like hybrid CNN-transformer search space with searchable downsampling positions;

(3) Our searched hybrid CNN-transformer models achieve up to 82.2% accuracy on ImageNet, surpassing EfficientNet by 2.1% with comparable compute time.

[Xiaojun Chang @cxj273]

@xwang_lk The official account says only 6432 active submissions.

[Xiaojun Chang @cxj273]

Our survey paper on Neural Architecture Search has recently been accepted by ACM Computing Surveys.

[arxiv @arxiv_org]

A Comprehensive Survey of Neural Architecture Search: Challenges and Solutions. https://t.co/7utcfsNlku

[Xiaojun Chang @cxj273]

@ShellyR79524350 @realDonaldTrump BY A LOT!

[Xiaojun Chang @cxj273]

@TobyWalsh PA will also not let Biden down...

[Xiaojun Chang @cxj273]

@poketpair @CharlesMcDoodle @adams2011 @ProjectLincoln @realDonaldTrump You are not following Trump’s “smart” suggestion, my bro. During the pandemic, the president has expressed confidence that the virus would quickly go away and skepticism over the effectiveness of face coverings.

[arXiv in review @arxiv_in_review]

#NeurIPS2020 Auxiliary Signal-Guided Knowledge Encoder-Decoder for Medical Report Generation. (arXiv:2006.03744v1 [cs\.CV]) https://t.co/pVB8oiYxDu

[Xiaojun Chang @cxj273]

‘Normal’ life is on its way.

[Victorian Department of Health @VicGovDH]

#COVID19VicData: Yesterday there were 11 new cases & the loss of 2 lives reported. Our thoughts are with all affected.
The 14 day rolling average & number of cases with unknown source are down from yesterday as we move toward COVID Normal. Info https://t.co/eTputEZdhs #COVID19Vic

[Xiaojun Chang @cxj273]

@rsh_bioinfo @PortezF @DanielAndrewsMP If you want to get back to normal as soon as possible, wear a mask!!!

[Junjie Hu @JunjieHu12]

Our recent work on Multimodal NMT is presented at ACL. Joint work w/ Po-Yao, @cxj273, Alex. We show that visual info can be used as a pivot source to improve NMT, in addition to be integrated as features in prior studies. https://t.co/GvaHIDIYE9

[Monash Information Technology @MonashInfotech]

Covered by the @australian, @MonashUni researchers have helped develop an advanced medical report generation method using #AI tech. This will accurately read lung CT images during testing – enhancing #COVID-19 diagnoses.

https://t.co/CruJtZ8Ca9
@monashdigital #socialgood #IT