chriskanan

10 ms·

61.
by chriskanan 2y ago
The main pubmed website seems to be working for me as well and I'm in upstate NY: https://pubmed.ncbi.nlm.nih.gov/ NIH/NSF pages have always gone down periodically for maintenance or otherwise, where this tends to
62.
by chriskanan 2y ago
I'm using PubMed to prepare for a lecture I'm giving on Monday right now and it is working fine: https://pubmed.ncbi.nlm.nih.gov/
63.
Universities Must Embrace AI or Face Extinction(chriskanan.com)
2 points by chriskanan 2y ago|0 comments
64.
by chriskanan 2y ago
I wrote this post because I don't hear anybody talking about this or planning for it. I've been telling people, including university administrators, about this issue for the past two years, and while my arguments seem to resonate,
65.
AI and the Existential Crisis Facing Higher Education(chriskanan.com)
4 points by chriskanan 2y ago|1 comments
66.
by chriskanan 2y ago
This will have a really big impact on NIH research at academic hospitals and a much greater impact on non-profit biomedical research institutes (e.g., Salk). The way these grants work varies across Federal granting agencies, but generally e
67.
NIH fixes indirect rates on new grants to 15%(grants.nih.gov)
8 points by chriskanan 2y ago|6 comments
68.
by chriskanan 2y ago
That's a good point regarding the Matrix Cookbook. I didn't start using it until I'd already learned a lot of linear algebra and how to do matrix calculus by hand, so I didn't really need much of an explanation when I fi
69.
by chriskanan 2y ago
The Matrix Cookbook was one of the most useful texts I had in the late 2000s during my PhD for computing update rules for loss functions. With autograd now widely employed, I don't think it is as fundamental as it was. Given that, what
70.
by chriskanan 2y ago
This paper motivates and explains concepts much better than the Tensor Cookbook.
71.
by chriskanan 2y ago
A recent Major Research Instrumentation proposal that I submitted to NSF had this required section in the Broader Impacts section [1], where I've pasted the text from the instructions: Institutional Commitment to Diversity and Inclusio
72.
by chriskanan 2y ago
I'm not sure why the link no longer works, but this one works. The link should be updated to this one: https://www.federalregister.gov/documents/2025/01/15/2025-00...
73.
by chriskanan 2y ago
I have no idea if comments actually have any impact, but here is the comment I left on the document: I am Christopher Kanan, a professor and AI researcher at the University of Rochester with over 20 years of experience in artificial intelli
74.
by chriskanan 2y ago
The most salient thing in the document is that it put export controls on releasing the weights of models trained with 10^26 operations. While there may be some errors in my math, I think that corresponds to training a model with over 70,000
75.
Framework for Artificial Intelligence Diffusion(federalregister.gov)
154 points by chriskanan 2y ago|130 comments
76.
by chriskanan 2y ago
Getting into a PhD program requires a lot of preparation soon after entering college. Having a parent with a PhD probably helps a lot to guide their child towards the necessary steps to get into a good program. I was totally clueless and th
77.
Publishers are selling papers to train AIs – and making millions of dollars(nature.com)
1 points by chriskanan 2y ago|1 comments
78.
by chriskanan 2y ago
I don't understand why in the current era we don't have videos just post-processed by the media player / TV. That seems like it would increase accessibility while not irritating folks who do not have epilepsy. I tried to sear
79.
by chriskanan 2y ago
On the flip side, potentially very good treatments that can't be patented or won't yield a good profit margin are not explored because it is very expensive and time consuming to get approval for a drug or treatment.
80.
by chriskanan 2y ago
This is what they would like to do, but it is hard to get funding for such an effort or to prioritize such work. It sounds like they think it would take a long time, and it wouldn't yield high impact papers from just doing that exercis
81.
by chriskanan 2y ago
Based on what the physicists have told me, the main issue with the Fortran code is that it is SLOW and hard to accelerate. Because the codebase is 30+ years old, much it sounds like it isn't well understood. They think it would benefit
82.
by chriskanan 2y ago
There is a lot of AI research in the nuclear fusion space. For inertial confinement fusion (a competing technology to magnetic confinement fusion, e.g., tokamaks) the National Ignition Facility (NIF) used it for their experiment that result
83.
by chriskanan 2y ago
Of the people who are still alive, Hopfield and Hinton make sense. Hopfield networks led to Boltzmann machines. Deep learning started with showing that deep neural networks were viable in Hinton's 2006 Science paper, where he showed th
84.
by chriskanan 2y ago
Here is the reasoning: https://www.nobelprize.org/uploads/2024/09/advanced-physicsp... I'm surprised Terry Sejnowski isn't included, considering it seems to be for Hopfield Nets and Boltzmann machin
85.
by chriskanan 2y ago
The assessments of visual capability really need to be more robust. They are still using datasets like VQAv2, which while providing some insight, have many issues. There are many newer datasets that serve as much more robust tests and that
86.
by chriskanan 2y ago
AI professor here. Most of AI is much closer to engineering or our knowledge of electricity before Maxwell, where we had been using it for 100+ years with big gaps in our understanding. Statistical learning theory is the branch of AI that i
87.
by chriskanan 2y ago
This study is great and addresses a question I've had about LoRA for a while. In a continual learning paper from last year, I found LoRA was extremely effective for faster fine-tuning and not forgetting the original dataset: https:&#x
88.
by chriskanan 2y ago
Exactly. Their brand is tightly interwoven with the animatronic band, but they haven't been evolving that. With generative AI and improved robotics they could create much more immersive experiences. They are damaging their brand. Kids
89.
by chriskanan 2y ago
This is great news. As an academic, I'm aware of multiple labs that built boxes with 4090s, not realizing that Nvidia had impaired P2P communication among cards. It's one of the reasons I didn't buy 4090s, despite them being
90.
by chriskanan 2y ago
I'm guessing they will then lobby to ban open source AI models, especially their usage in commercial applications. One would need a huge bank account to create models in the future. I think this would have a chilling effect on the AI e

More ›