What's on Your Mind

Page 1698 of 3729
Reply
  • Aug 8, 2022
    ·
    2 replies
    fiveprestos

    seems intriguing - anywhere u can recommend I learn more?

    hmmm idk tbh like id say read some papers but they cost a lot to read unless u have a institution u can log in thru

    i have some texts on machine learning i can share, thats the bulk of what i do. the image processing stuff is pretty limited in scope but i could also discuss how it works etc

  • Semi 🐬
    Aug 8, 2022
    ·
    1 reply
    Shammy

    hmmm idk tbh like id say read some papers but they cost a lot to read unless u have a institution u can log in thru

    i have some texts on machine learning i can share, thats the bulk of what i do. the image processing stuff is pretty limited in scope but i could also discuss how it works etc

    Also interested

  • Aug 8, 2022
    Shammy

    hmmm idk tbh like id say read some papers but they cost a lot to read unless u have a institution u can log in thru

    i have some texts on machine learning i can share, thats the bulk of what i do. the image processing stuff is pretty limited in scope but i could also discuss how it works etc

    Yeah I'm all ears. Feel free to dm me just lmk when sent. All greatly appreciated

  • Aug 8, 2022
    ·
    1 reply
    Semi

    Also interested

    ill try to explain in short what im doing @fiveprestos

    I have 2 types of images. images of cell nuclei (which we use to identify individual cells) and images that indicate the activity levels of a particular protein.

    We need to first identify nuclei in images, so to do this i use this software called ilastik. Here you basically color in the pixels that correspond to nuclei (think MS paint or photoshop paint brush etc), and the software can then automatically identify which pixels correspond to nuclei.

    The cool part is the way the software does this is it extracts "features", or basically just numbers/data derived from different formulas that tell you information about the properties of the image. The machine learning part comes into play when you use these "features" to classify pixels as either nuclei or background. There are many different machine learning models u can use (logistic regression, SVM, neural networks) but ilastik uses whats called a random forest. That is a very useful algorithm because it doesnt require a ton of data to be accurate, but it can still predict accurately with data that has nonlinear behavior (basically the relationships between the data points are very complicated)

    Once we have the segmented, or individually identified nuclei, we just measure the activity level of a protein of interest on a per-cell basis. In practice we also track the nuclei and measure the protein activity over time on a percell basis. We can then correlate the protein activity with various cell behaviors like growth, division, migration etc.

  • CKL TML 🌺
    Aug 8, 2022
    ·
    1 reply

    Man what a terrible day
    Glad i got a nightshift so i can dip from this crib

  • Aug 9, 2022

    im 5'10+

  • CKL TML 🌺
    Aug 9, 2022
    CKL TML

    Man what a terrible day
    Glad i got a nightshift so i can dip from this crib

    Time flew by

  • CKL TML 🌺
    Aug 9, 2022

    Imma get a haircut today
    That might fix my moodiness

  • Semi 🐬
    Aug 9, 2022
    Shammy

    ill try to explain in short what im doing @fiveprestos

    I have 2 types of images. images of cell nuclei (which we use to identify individual cells) and images that indicate the activity levels of a particular protein.

    We need to first identify nuclei in images, so to do this i use this software called ilastik. Here you basically color in the pixels that correspond to nuclei (think MS paint or photoshop paint brush etc), and the software can then automatically identify which pixels correspond to nuclei.

    The cool part is the way the software does this is it extracts "features", or basically just numbers/data derived from different formulas that tell you information about the properties of the image. The machine learning part comes into play when you use these "features" to classify pixels as either nuclei or background. There are many different machine learning models u can use (logistic regression, SVM, neural networks) but ilastik uses whats called a random forest. That is a very useful algorithm because it doesnt require a ton of data to be accurate, but it can still predict accurately with data that has nonlinear behavior (basically the relationships between the data points are very complicated)

    Once we have the segmented, or individually identified nuclei, we just measure the activity level of a protein of interest on a per-cell basis. In practice we also track the nuclei and measure the protein activity over time on a percell basis. We can then correlate the protein activity with various cell behaviors like growth, division, migration etc.

    Fun, sounds rewarding

  • Semi 🐬
    Aug 9, 2022

    Energy in the universe feels strange

    Fever dream esque

  • Aug 9, 2022
    ·
    1 reply

    not that long ago social media was on Pardi's d***. as of yesterday they've done a full 180 and won't stop slandering that man.

    yeah, it's corny to take a petty joke to heart... who wouldn't wanna be meg's pet? but the way they're on this niggas head is wild. sending love to whoever is a public figure it ain't healthy out here

  • the quicker you face it, the quicker it's over

  • mishima 😈
    Aug 9, 2022

    My Smash Bros Ultimate friend doesn’t wanna play with me anymore because I’ve nearly mastered Kazuya :(

  • Aug 9, 2022
    ·
    1 reply

    good news good news good news thats all they wanna hear

  • Aug 9, 2022
    sword

    not that long ago social media was on Pardi's d***. as of yesterday they've done a full 180 and won't stop slandering that man.

    yeah, it's corny to take a petty joke to heart... who wouldn't wanna be meg's pet? but the way they're on this niggas head is wild. sending love to whoever is a public figure it ain't healthy out here

    it's not easy. I feel like you gotta detach from the internet's approval. don't seem them as your ally when they big you up because they will turn on anyone at anytime. you should seek approval/love from real friends and fam in your life

  • Aug 9, 2022
    PainPapi

    good news good news good news thats all they wanna hear

  • someone gotta make a traffic thread like how theres a work thread

  • Semi 🐬
    Aug 9, 2022

    I'm not good at closing sales, but I can get someone to that close point.

    I'm also bad at actually providing my skill set and knowledge to the degree that I can, because I think I assume people know things that they don't so I don't give enough information and it may come off as condescending/patronizing.

    I forget that I actually have specialized knowledge, and that it is specialized so when people are asking me questions I should answer using that knowledge.

    I also just have this underlying feeling that Im wrong but I need to let that go, because I do know what I'm talking about.

  • Aug 9, 2022
    ·
    1 reply

    Just scrolled on tiktok for 3 hours I want to kill myaelf

  • some white ppl really doesn't understand why they don't say the nword that's why they get caught saying some weird s***

  • Aug 9, 2022
    PainPapi

    Just scrolled on tiktok for 3 hours I want to kill myaelf

    Had to delete that. Always fell into an algorithm that makes me feel like s***

  • Aug 9, 2022

    I remember when I was a kid, I'd wake up and turn on MTV because they'd have the time in the corner so I can keep track while I got ready for school.

    They'd play the morning music videos, usually a rock or pop song.

    Think that had one of the biggest influences on my music wise to venture outside of hip hop. Still remember a lot of those same songs.

  • Aug 9, 2022
    ·
    1 reply

    I’m so sick of my music stopping when I’m browsing other apps idk what iOS update got rolled out but it’s f***ed my vibe up

  • Aug 9, 2022
    ·
    edited
    ·
    1 reply

    I find it extremely difficult to enjoy life.