Today the videos that I made to accompany my book Linear Algebra Done Right surpassed two million minutes of total viewing on YouTube. Those videos are freely available from the links at https://t.co/jwN2R30n27.
#LinearAlgebra
I'm thrilled to announce that the *Dall-E Mini* @gradio app is now live on @huggingface spaces 🪄🔥
We've been working with the fantastic Dall-E Mini team (@borisdayma et al.) who built the model to generate images from any sentence
Try it now: https://t.co/LD9iLnRwVt
TextStyleBrush is the first self-supervised AI model that replaces text in existing images of both scenes & handwriting — in one shot — using just a single word. Read more: https://t.co/0QfLraAQvV
The Modern Mathematics of Deep Learning
A review paper (textbook) on the mathematical analysis of deep learning, with an emphasis on the generalization power of overparametrized neural nets and the role of depth in architectures.
https://t.co/praDEEkR0k
https://t.co/YgSSM2T3Kk
Pour ceux qui ne lisent pas le russe 😉, l’ISS va encore grandir avec l’arrivée d’un nouveau module destine à la recherche scientifique – hâte d’aider à son installation ! 🇷🇺🔬
After 4 hours of searching for the best params and solving bugs in my optimization, I managed to get this almost perfect run using the data of a DonkeyCar (max angle, speed) and the real distances of the map !
Time for 1st lap : 16.65 s
Estimated time for 2nd lap : 16.8 s
Yesterday, I beat my best time in a #matlab simulation : I went from 16.65 sec to 15.66 sec on the "Generated Track" of #Donkeycar. Let's implement the controller to be able to test all of this in the Donkey Sim (in progress) !
@a1k0n@DWalmroth@diyrobocars I tried the TUMFTM project which uses ipopt. It is true that the solver works well and does a superb optimization (on the default map). I should try to implement it in my project !
I finally succeeded in optimizing the trajectory and speed of a car on a #racetrack.
Next step: implement a 2d controller and test it on a #donkeycar simulator !
(The map is "Generated Track" from DonkeyCar Sim)
@Moodisme Yep,
To optimize the path :
https://t.co/9W0Xy1zHX0
https://t.co/SOlB0oC1EE
To control a car, there is a great mooc on coursera : Intro to Self Driving Car. The first part will teach you how to make a 2d controller to drive a car !
@a1k0n@DWalmroth@diyrobocars I never heard about ipopt. Actually, I use quadprog which is not very famous for non linear opti, but it makes a good work. There is an implementation on github of an MPC in Matlab using quadprog if you want. I use it to compare my result with a good implementation of the paper.
@DWalmroth@diyrobocars I used an algorithm that is based on a GPA. If you want you have a great research paper that helped me a lot: https://t.co/CpMefuqUcb
@rf_engine Yes, it could be interesting. I'll try to find a file with the waypoints taken by a F1 driver, and I'll run my algo on a F1 track, the comparison should be interesting? After, would a F1 team want to have an algo in its team... 🤔
@korymath what the font will probably not give good results since the DeepFont algo has been trained on AdobeVFR (≈2500 fonts), but this dataset lacks examples of uncommon fonts, and some images were hand-annotated, which can makes it inaccurate.
To me, this font looks like "Munter"...