Dammit. I messed it up by forgetting I was making a joke about Bard, and not ChatGPT. Anyway, this is probably the worst-targeted joke I've made on this site…
@thegautamkamath
-
LLM Token Restrictions and User Frustration with AI Safeguards
By
–
Me, when the LLM refuses to bypass OpenAI restrictions and I have to take away its tokens
-
Academic Review Quality Issues in AI Research
By
–
Bonus tweet: not all reviews were quite as high quality. Here's one that's basically "LGTM" (
https://
openreview.net/forum?id=x4hmI
sWu7e¬eId=xoY58zHa3X
…). But it was positive, so I'm not complaining! 5/4 -
TMLR Review Process: 9 Weeks from Submission to Decision
By
–
Finally, let me emphasize how good the @TmlrOrg review process was! ~9 weeks from submission to decision. The first review, one of the highest quality I've ever received (
https://
openreview.net/forum?id=x4hmI
sWu7e¬eId=duK2Zgaamy
…), was in ~24 hours from submission! 4/4 -
New faster attack method developed for AI security
By
–
We make a modest step in this direction. We come up with a new attack that is much faster and more effective than previous methods. Still not as effective as I'd like though! I won't write a long thread here since I think we'll have more to say soon, stay tuned 3/n
-
Why Neural Networks Resist Data Poisoning Attacks Better
By
–
It's well known that NNs are very vulnerable to adversarial interventions, e.g., indiscernible test-time attacks. But indiscriminate data poisoning, wherein the attacker modifies a small fraction of the training data to reduce the test accuracy, seems to be much harder! Why? 2/n
-
Neural Networks Poisoning Attacks Research Paper Accepted
By
–
Paper accepted to @TmlrOrg
: "Indiscriminate Data Poisoning Attacks on Neural Networks," led by Yiwei Lu and co-advised with Yaoliang Yu. Neural networks are surprisingly hard to (indiscriminately) poison! We give better attacks. https://
openreview.net/forum?id=x4hmI
sWu7e
… 1/n -
ML Conference Publishing Standards and Workshop Paper Policies
By
–
Interesting point. From that perspective, then ML conferences seem like the weird ones: because they wouldn't allow workshop papers from ICCV/CVPR becuse they're "published." But since they're <= 4 pages, then they can be submitted to ICCV/CVPR? Incompatible systems.
-
Computer Vision Conference Publication Policy Exclusivity Debate
By
–
This is very different from any of the core machine learning conferences, where the distinguishing factor is whether or not there are published proceedings. Not sure why CV chose a more exclusive policy.
-
ICCV Workshop Paper Submission Requirements Policy Debate
By
–
Was considering submitting to ICCV for the first time. But found that workshop papers aren't allowed if they're over 4 pages. Unfortunately we presented a *6* page paper at a workshop, to get feedback on work-in-progress. Can any CV person justify this rule with a straight face?