HiPerGator Symposium - Shared screen with speaker view
Who can see your viewing activity?
Okay...after watching this video, how many people are totally grossed out and dread your next trip to Publix?!?!
Jorge X. Santiago
Adrian roitberg ipad
how are you going to deal with the obvious bias danger in this project? who decides what is a hazard? you must know this will end up with people of color being targeted, right?
That was SUPER Gross!!!!!
If being captured, do they resolve the obvious issues?
No wonder the packages are SEALED!
this interdisciplinary approach to implementation is amazing. the simulation capability is fantastic.
How can AI predict human behavior if developed with bias algorithms? This project reminds me of AI in Minority Report to predict crimes in advance.
It seems as if there is some confusion - the HazardNet project isn’t attempting to predict behavior, we are attempting to detect coughing in retail stores so that employees can respond and clean up the area or provide people with masks.
Adrian roitberg ipad
you explicitly mentioned police! come on....
We have no data indicating any racial group commits more tampering or aggression than another; and a very diverse group of participants are involved in creating the dataset to train possible models. The model will ID possible tampering with or aggressing against others. We're working to protect victims.
My mistake. Thanks for clarification. Pardon.
this agricultural food based approach could be very useful within the context of GIS mapping and health issues.
definitely a good approach to generate 'simulated' data; the researchers can 'control' participants and avoid detection biases associated with race or other factors. Overall nice approach!
Yes! A good model
Community members serving as police are sometimes called by fellow citizens to protect place users against intimidating and dangerous behavior including intentional infection threats and actions.
Rabin, very interesting data, thank you for presenting.
Very informing talk, Rabin. Thank you!
Very informative and interesting talk Rabin, Thanks
great way to make data less dirty within data input. it’s awesome.
wow, these data sets overlaying with one another in conjunction with the land grant status/extension system/health system. UF is quite a powerful place to roll out some integrated real world solutions for health, climate, and equitable economic development. And, the academic/public nature in conjunction with sunshine laws really assures safety measures remain in place. it would be great to help. (ADRC systems, clinical, govt, etc). neat stuff.
Jorge X. Santiago
interesting data set
Really interesting work, Vincent! Thank you for sharing with us.
Great work Vincent!
Really great content, Vincent!
Great title Dr. Davis and Dr. Roitberg
so... neural nets can theoretically learn any function composition... super cool idea to teach a NN quantum mechanics !
Nice introduction of ANI and pretty cool future work, Kate!
we need full protein models!! :)
that’s super cool, In could see how those quantum data points inputted could parlay answers in other complex fields.
wow tremendous capabilities for shortening the life cycle of vaccines and end meds. the correlation from stucture to purpose makes a clear line to clinical study. you could overlay EMR end points and search for anomalies... really breathtaking.
you could quickly revisit and expand disease state cross over with pleotrooic effects.
Spectral Energy Distribution for those that don't know
@Bryan, great talk quick question about your systems, how do you deal with merging the different intermediate representations in the layers close to the output? doesn't your network get very complex as depth increases?
@ Alexander Kane sorry for missing your question. it is ~ 1kcal/mol level of accuracy compared to reference level of theory
super cool, climate energy capability and home insurance premiums 🥰
@ Joshua miles there are many actually, and a lot are doing some similar things to the work I am trying to accomplish. People are doing things with charges and long range interactions. Physnet, which is energies, forces, and mulitpoles I think. These are just a couple of examples
amazing really, it gets into super conductors and availability of energy transfer amid power grids and available streams of energy, etc. amazing work.
@Ignacio - reasoning modules get all the intermediate features; the depth of the feature space does grow linearly with # of network layers. Avoiding dense neural layers helps control parameter count.
there are buoy systems which could benefit from this...
Why is by contradiction the most likely/prevalent? Is there an understood reason for that?
this language sounds a lot like economic theory of equitable development.