William Thong (@twuilliam) profile picture

Machine Learning @Apple. Prev: Responsible AI @Sony, PhD @UvA_Amsterdam, MSc @polymtl/@Mila_Quebec. Views my own.

Zurich, Switzerland
Joined August 2010
Photo by William Thong (@twuilliam)Photo by William Thong (@twuilliam)Photo by William Thong (@twuilliam)Photo by William Thong (@twuilliam)Photo by William Thong (@twuilliam)Photo by William Thong (@twuilliam)
We made it to @Nature πŸŽ‰πŸŽ‰ FHIBE establishes new ethical standards in data collection and raises the bar for fairness benchmarking of human-centric computer vision models. A massive congrats to the whole team! Very proud of this achievement done during my tenure @SonyAI_global
1
297
Personal update: I have recently joined @Apple! Very excited to see what we'll achieve together! I am beyond thankful for my enriching 3 years at @SonyAI_global, where I had the opportunity to lead a team of researchers and improve practices in responsible AI.
2
1
47
4,001
Our paper raises ethical considerations in visual data collection, and offers perspectives on how to think about them from the start. Work led by Jerone Andrews! With @dorazhao9 , myself, @amodas_ ,@SciOrestis ,@alicexiang Paper: arxiv.org/abs/2302.03629 #NeurIPS2023
The Sony AI paper, β€œEthical Considerations for Responsible Data Curation,” offers proactive, domain-specific recommendations covering purpose, privacy, and consent, as well as diversity, for curating human-centric computer vision (HCCV) evaluation datasets. #NeurIPS2023
7
1
20
2,852
Go read @peard33 's great article in @WIRED ! It gathers the perspective of several scholars on skin color bias in computer vision. Thank you for the valuable coverage of our #ICCV2023 paper.
By expressing skin tone using only a sliding scale from lightest to darkest or white to black, today’s common measures of AI effectiveness ignore the contribution of yellow and red hues to the range of human skin. πŸ“· Courtesy of Sony // πŸ”— wired.trib.al/iarGnTt
6
550
Interviewing with @Melissahei was a real pleasure! Thank you for the coverage of our #ICCV2023 paper in @techreview . Hope this will foster further awareness and discussion about skin color bias in computer vision!
Computer vision systems are everywhere. But they are riddled with biases, and they’re less accurate when the images show Black or brown people and women. Two new papers propose ways to measure biases to more fully capture the rich diversity of humanity. trib.al/We9VigA
1
3
6
1,505
And many thanks to Guha Balakrishnan for the kind words about our work!
144
See you next week at #ICCV2023!
Introducing β€œBeyond Skin Tone: A Multidimensional Measure of Apparent Skin Color,” a new solution for bias identification and assessment by Sony AI that takes a significant step towards a more comprehensive understanding of skin color in #computervision. ow.ly/byaB50PPjKI
1
14
795
Excited to announce our new @FAccTConference paper, β€œAugmented Datasheets for Speech Datasets and Ethical Decision-Makingβ€œ, led by @SciOrestis and @anna_sg_choi (who are presenting today) & @alicexiang! Link: dl.acm.org/doi/10.1145/35930…; πŸ§΅πŸ‘‡(1/9)
1
21
3
78
23,740
Interested in image quality assessment? Or how to derive differentiable correlation coefficients? Join us today at #BMVC2022 to know more! Paper: arxiv.org/abs/2211.05215 Joint work with Jose Costa Pereira, @s_parisot, Ales Leonardis and @steve_mcdonagh
3
5
William Thong retweeted
3, 2, 1 🏁 Meet Gran Turismo Sophy, a superhuman racing AI agent. Developed as a collab between Sony AI, PDI, and SIE, #GTSophy is the world's first AI agent to outrace the world's top players in @thegranturismo, achieving a new breakthrough in #AI. ow.ly/F2sr50HPQuS
13
172
37
793
William Thong retweeted
What is #GTSophy? 🏁 GT Sophy is an autonomous #AI agent trained utilizing a novel deep reinforcement learning platform. Our breakthrough research "Outracing Champion Gran Turismo Drivers with Deep Reinforcement Learning" was published today in @Nature. ow.ly/4RKH50HQOWt
4
95
25
274
Honoured to see my name among the outstanding reviewers of #BMVC2021 !
🍾 Well done to the #BMVC2021 outstanding reviewers!
10
Excited to share that my first work in the field of algorithmic fairness and bias mitigation has been accepted at #BMVC2021! Come say hi if you are attending the conference πŸ€— 🧡 (1/4):
#BMVC 2021: Feature and Label Embedding Spaces Matter in Addressing Image Classifier Bias w/@twuilliam TL;DR addressing #image #classifier #bias on #feature and #label spaces #improves prediction #fairness, while preserving classification performance isis-data.science.uva.nl/cgm… 1/n
1
9
We are also grateful to the work of Zeyu Wang and colleagues at @VisualAILab (CVPR'20) for benchmarking multiple mitigation methods. On these benchmarks, we improve the fairness of image classifiers by switching to a label embedding space, instead of a one-hot encoding. (3/4)
1
In summary, we derive real-valued vector representations for every class and each protected attribute value. By doing so at both feature and label spaces, we can improve algorithm fairness and preserve classification performance. (4/4)
Glad to see that my reviews were helpful! 😊
Congrats to all the #CVPR2021 Outstanding Reviewers! Source: cvpr2021.thecvf.com/node/184
7
Interested in zero-shot action recognition with literally *zero* training video? Here's our recipeπŸ‘‡ 1. Identify what objects are present. 2. Derive spatial and semantic priors to enhance the matching between objects and actions!
Learning from examples is fun, but can you localise and classify actions without having seen a video before? We find that with the right prior spatial and semantic knowledge from objects, more is possible than you think. In IJCV w/ @twuilliam @cgmsnoek arxiv.org/abs/2104.04715
5