Someone

Building ML models & quant strategies | Math proofs & code breakdowns | Software dev turned trader | Nigeria 🇳🇬

[Someone @random_user_876]

I am a mathematician
Software Developer
And i follow jesus ❤️

[Someone @random_user_876]

@Isaaccarcacia This put a smile on my face, congrats! 🎉 🙌

[Isaac Carcacía @Isaaccarcacia]

Oficialmente doctor en matemáticas.

[Nihit Singh @xnishidh]

L1 & L2 loss functions

helps a ml model to reduce error and avoid overfitting, for regression we use loss functions

- l1 loss > mean absolute error (MAE)
- l2 loss > mean squared error (MSE)

for L1 loss function measures your absolute difference between the actual value and the predicted value

therfore the graph is linear for error and penalty
if the error comes to be 10 then the penalty is 10

>> error = penalty

hence big errors don'ts get excessive penalties. In L1 every unit of error is equally bad

This is also called Mean Absolute Error(MAE).

for L2 loss function measure the overall discrepancy between a models predicted values and the actual target value by summing up the error before average

therfore the graph is linear for error and penalty
if the error comes to be 10 then the penalty is 100

>> (error)^2 = penalty

hence it reduces large error firsts by making adjustments

During training

model make predictions
|
computes the absolute error
|
use gradient descent to adjust weights to reduce errors

w = w − η × gradient

Total loss = Actual loss + penality

for L1 > Λ + sum( | Wi | )
for L1 > Λ + sum( (| | W | |i )^2 )

L1 is more resistant to outliers, while L2 is more sensitive to them

[Martin @martinmrmar]

When you decide to be a math major to study pure mathematics, these are some of the math courses you will be encountering. It's been over 20 years for me since grad school days and I still remember the fun times in those classes. I focused on Real/Complex analysis, Topology, and

[Someone @random_user_876]

@mujib1612 Ragna Crimson

[Someone @random_user_876]

So people actually follow these laws 🤦‍♂️

[Stormi @Stormi_luv]

NO cheating, post your last saved image without any context 💀💀

[Someone @random_user_876]

@elmustyy_ Scholarrrrr 🙌

[EL Mustapha @elmustyy_]

Another reason not to drop off the track and continue with the pace .
A list just dropped at the department containing all first-class students across all levels ,
I spotted my name 🫣 .
It's too early to be proud, honestly ..
I hope the energy continues .

[EL Mustapha @elmustyy_]

I'll be so good that I'll amaze myself ...

[Someone @random_user_876]

Multiplying by 3 shouldn't cause a crisis of faith, yet here we are.
Let the $0.999... = 1 debate begin.

[Someone @random_user_876]

Euler was a true genius

[Someone @random_user_876]

@zcahverse

[UI/UX Savior @UiSavior]

🔥

[Kaal Drogo @sykquill]

@martinmrmar One proof by Taylor series expansion, another by Euler's formula

[Someone @random_user_876]

@elmustyy_ 😪

[Someone @random_user_876]

Seriously, why doesn't anyone talk about 270 degrees?

[Roro ☃ @RoRoFli]

ChatGTP: Just say go, and I will generate a realistic HD image

Me: Go

ChatGPT: You have reached your free limit