One Reason You May Struggle To Learn ML/AI
ฝัง
- เผยแพร่เมื่อ 22 พ.ย. 2024
- This is one of the main reasons people struggle to get into the machine learning and artificial intelligence field!
Watch the full video here: • Should You Learn Machi...
💻 Master Blockchain and Web 3.0 development today by using BlockchainExpert: algoexpert.io/... - use code "tim" for a discount!
💻 ProgrammingExpert is the best platform to learn how to code and become a software engineer as fast as possible! programmingexp... - use code "tim" for a discount!
◼️◼️◼️◼️◼️◼️◼️◼️◼️◼️◼️◼️◼️◼️
👕 Merchandise: teespring.com/...
📸 Instagram: / tech_with_tim
📱 Twitter: / techwithtimm
⭐ Discord: / discord
📝 LinkedIn: / tim-ruscica-82631b179
🌎 Website: techwithtim.net
📂 GitHub: github.com/tec...
🔊 Podcast: anchor.fm/tech...
🎬 My TH-cam Gear: www.techwithti...
💵 One-Time Donations: www.paypal.com...
💰 Patreon: / techwithtim
◼️◼️◼️◼️◼️◼️◼️◼️◼️◼️◼️◼️◼️◼️
⭐️ Tags ⭐️
Tech With Tim
Machine Learning
⭐️ Hashtags ⭐️
#shorts #techwithtim
As a Data Scientist I can confirm that it can be very frustrating. Unlike with software engineering, with ML you never know what you're gonna get. Maybe you spend a ton of time coding, cleaning data, and engineering features only to find out that your data simply doesn't allow you to build a sufficiently accurate model. In contrast, in software engineering, you'll always have something to show if you put in enough work.
I see ML/AI one of the fields where really university can help you understand. Because math is really crucial on these concepts.
That’s good to hear! I’m starting my AI bachelor in september
Yes it's simply way more theory based, and way less implementation oriented, unlike software engineering
I totally agree. I have a degree in applied mathematics and if I didn’t I’d probably be lost.
@@brendanlydon5272 do you really need anything beyond linear algebra and statistics?
@@Milark a little calculus too, but mainly linear algebra. To really understand these systems though is knowing that all these concepts are intertwined. Everything in life and ML/AI is math though, so that made the decision easier for me in terms of what I wanted to study. I don't think its worth going to school for computer science or even "AI" especially when all these systems are based on deep mathematical principles and information is so readily available for learning core programming concepts. It does you no good not knowing what or why your programming, and I think thats the big draw back of these higher level libraries/languages and what Tim is saying. Everything is being abstracted away from the programmer with things like keras, tensorflow, etc. So, ya you can go to school for AI or CS all you want, but if you cant even solve for a line of best fit using the least squares method mathematically without calling np.polyfit() are you even a "professional" ML/AI developer? If you want to get to that "professional" level you have to be excited about solving the problems that require no programming. Programming is just a vehicle for conveying your ideas not the ideas themselves.
One reason that struggling my ML journey is because my device didn't strong enough to run it 😅
there are cloud services to train insane models
Of course brother every activities need focus and time
linear algebra and gradient descent are high school level arent they? /gen, ofc most ppl wont have it in the back of their minds but a quick refresh course will remind you of the basics, calculus and algebra needed for ml is actually pretty simple
Gradient descent involves very simple math
I wish someone had told me long ago, that being a passable software engineer didn't require expert math, I would've been more open to learning the science long ago
@techwithtim is AI ML hard for a beginner to understand or study
Could u make Algo videos or python all DS
Good info
Not really, the maths in ML are quite limited and simple, and most of it are 100% taken care by simple blackbox functions and tensor specialized HW. The understanding, analytical side and innovative part is the hardest part in ML, the math part is trivial even compared to high school maths, and mostly skippable
I would argue though that understanding what is going on under the hood can help you gain a better understanding
@@barni_7762 Understanding the maths in ML are extremely simple. It is extremely straightforward (that is why it is completely delegated to the GPU) just arduous. Coming up with new use cases, understanding data quality, deciding on desired input and output scopes and syntax, THAT is what makes ML hard, not the math
I'm myself trying to get into ML a bit. I totally agree. It's simple highschool calculus. Still, though, writing my own small autograd engine helped me understand problems like gradient vanishing better. Of course what you mentioned requires a deeper understanding of the topic, there's no doubt about that.
If you wanna optimize your backprop a non-trivial the problem you're trying to solve. You're gunna need to know vector calculus
@@seventeeen29 I dont know what youre trying to say exactly but it is built for numpy arrays, so it isn't scalar valued (otherwise it would just be micrograd)
learning ai yeah lmao
Nah with AI in its current stay it's not hard