Shasha Li

PhD candidate @UCRiverside. Research on machine learning and security.

[Salman Asif @salmanasif]

Happy to share our paper on context-aware adversarial attacks, recently accepted to #AAAI2022

Joint work with amazing students @ZikuiCai, Xinxin, and collaborators @VCG_UCR, @laosong, Srikanth, @ShashaLi16, and Mingjun.
Congratulations everyone!

https://t.co/nWKGUcN8ox

[Shasha Li @ShashaLi16]

Delighted to share that our paper(https://t.co/yie7S33Mrb) is accepted by #NeurIPS2021. Your video classifier is openly vulnerable to simple geometrically transformed perturbations! Thanks to the great team @aaich001 @zst_rising88 @laosong (and others not on here). #adversarialML

[NEC Labs America @NECLabsAmerica]

NEC Labs America is hiring Research Interns for Summer 2021! If you are interested in joining our team, apply here: https://t.co/5F9O82fn6H

[Shasha Li @ShashaLi16]

📢📢

[Shitong Zhu @zst_rising88]

📢📢 Excited to announce our paper titled "You Do (Not) Belong Here: Detecting DPI Evasion Attacks with Context Learning" has been accepted (and presented) at @ACMSIGCOMM
#CoNEXT2020! Big shout out to the great team @ShashaLi16 @zhongjie_wang_x @pkqzy888 (and others not on here)

[Shasha Li @ShashaLi16]

Thanks for the great summary👍

[Shitong Zhu @zst_rising88]

Preprint of our #ECCV2020 work on context-based defense against adversarial example: https://t.co/ya222ZOB5Z. Unlike prior work, our approach learns the "normal" context from clean images (e.g. stop sign very likely accompanies road crossing, not speed limits), and detects... 1/n

[Shasha Li @ShashaLi16]

Codes on the way!

[Chengyu Song @laosong]

Preprint: https://t.co/cQIo0P9SMh

[Shasha Li @ShashaLi16]

@zst_rising88 Exactly!

[Shasha Li @ShashaLi16]

Yep, our adversarial defense paper is accepted #ECCV! Context modelling is used to detect adversarial examples out of context, e.g., a speed limit sign (should be a stop sign) at a crossing road with stop line. Preprint and code coming soon.#adversarialML @laosong @zst_rising88

[Chengyu Song @laosong]

Got my first ECCV paper, on detecting adversarial perturbations using context-inconsistency, with our excellent students @ShashaLi16, @zst_rising88, and Sudipta.

[Shasha Li @ShashaLi16]

@DongkuanXu @neclabs Thanks, Dongkuan!

[Shasha Li @ShashaLi16]

Happy to announce that my intern work @neclabs is showing up in #InfoCom2020.
Attention mechanism is customized to deal with multipath propagation.
Looking forward to my very first online conference!

[Shasha Li @ShashaLi16]

https://t.co/8AdZoT08gh

[RUHS - Public Health @RUHSPH]

#Coronavirus update: We can confirm 185 #COVID-19 cases; 8 deaths in Riverside County. Visit https://t.co/B0PcBKTHe0 for an updated map featuring city, age, gender breakdown. We will update once a day in the afternoon going forward.

[Shasha Li @ShashaLi16]

TYPO is definitely the enemy of coding.

[Ram Shankar Siva Kumar @ram_ssk]

New Paper📢 : Adversarial Machine Learning - Industry perspectives

TL;DR:
- 25 out of 28 organizations we interviewed noted that they dont have right tools in place to secure ML assets
- SDL for industry grade ML models has lots of open questions

https://t.co/bxJS14xIpm 1/

[AI at Meta @AIatMeta]

We are welcoming Amazon Web Services and new academic experts to the Deepfake Detection Challenge, a collaborative effort to accelerate creation of new tools to detect manipulated videos and images.

https://t.co/oFjqAVnsng

[Shasha Li @ShashaLi16]

@Tesla 🤣😂🤣😂

[Shasha Li @ShashaLi16]

@ahmaddarkisc @acm_ccs Patpat...

[Shasha Li @ShashaLi16]

Adversarial Examples Are Not Bugs, They Are Features

[Ilya Sutskever @ilyasut]

https://t.co/Dl2VSXaX7O: a strange imagenet-like dataset with very wrong-looking labels, yet a model trained on it does totally well on the normal validation set.

It's a crime against ML!

[Shasha Li @ShashaLi16]

@pgbovine 🤣

[Shasha Li @ShashaLi16]

Interesting works #aistats2019 on Day1