Nobody thinks that, but at the same time, nobody wants to spend time writing a custom kernel, type management, and will simply opt for a python wrapper.
I learned x86 Assembly last year, it just blew my mind. It made my C++ code much cleaner and better and I feel I can write more optimized code than I would have without knowledge of Assembly language. Now plan is to bulid my own microprocessor RISC-V softcore on 20K LUTs FPGA and create my own ISA and assembly lanauge and after that make a C complier. Its going to be a fun ride 🤠
If you're not a beginner and you want to do machine Learning in c++ do it it going to be better and faster when you understand the stuff and don't want to use any library like me for example. I just want to write my own ml library in cuda and do my own models. If you want to do it... Do it
As someone who's proficient in both python and C++... i would prefer C++ any day. It's really interesting to code something new especially for ML projects
I wrote my interpretation of NEAT AI in C++. Wrote a few sample "games" that use this library. One of which is of tank bots that roll around and try to hit a moving target while not hitting friendlies. It works really well and processes much faster than an equivalent Python library. Now I'm writing up a Deep Q Learning library in C++ just because...
You are right about the hardest things for beginners are pointer/references, passing by x, classes and further on templates, smart pointers, function pointers, etc. But generally, I think C++ is the best language to start with, it is not as complex as people might think/say. It's like every (normal) programming language, high level languages are of course, extremely easy, but you lack the fundaments of the language by that. When learning C++, you need to know the fundamentals before you can begin, which I think is really good about the language. Later on in C++ you will learn about stl, the preprocessor directives, marco's, etc. So, I dont see any reason why people shouldn't be learning C++ as their first language.
Well, pragmatism would direct people to Python. There is no reason to tech an accountant who uses Excel C. People who are in the science fields don't need complex language features that C++ offers because they will never use them
@@TheBuilder There is no reason for an accountant to learn python. People in the science fields use applications that doesn't require programming skills.
You'll be surprised. In India, accountants in some firms need to learn a bit of python to do their work or they spend hours looking manually doing it often on an old PC monitor, not a laptop@@budgetarms. Hmmm why did the reference come here instead of the beginning, anyway, the python they learn is like a 1hr or 3hr crash course on simpler numpy and pandas rather than basics of python
Scikit-learn is mostly written in Python and Cython (not in C++), and thus it is not interchangeable with any other language. Neither sktime which is also written in Python.
there are some machine learning libraries built for c++ but its for in house only and not distributed publicly.. they actually use only for production.
Do you have any good learning sources though? I found "Geek For Geek - Machine learning with c++", a book called "Hands on Machine Learning with c++" and the usual library documentaries. But not to much else thats non-Phyton. Are there actually better places to start? Or paths that are usually taken? (My higher math is ok and c++ is getting there)
Interesting, would C++ be actually much faster than python in Machine Learning as well? I heard that overall c++ just kills python in speed, but does it also translate to much faster and power efficient ML? I would like to learn it if so
If you actually want to do machine learning though, your work will probably be using python and the open source you are building off of is probably also written in python
@@TheBuilderim a machine learning engineer and have never dealt with ML in any language other that python (with the exception of things like custom cuda kernels). I know there are positions that exist that use languages other than python for ML but they are just not as common, so it makes a lot of sense to me that resources would suggest the most commonly used language which also happens to be a very easy language
I am already writing a library for machine learning in C++
Give me
@@waldemir5621 you're going to be very cute and start by saying please.
and then you start by being less lazy S.O.B.
@thendimension4816, What about pytorch?
@@FILISPEEN good if you want a simpler syntax over more detailed one like tensor flow
@@thendimension4816 sorry, I confused pytorch with libtorch for c++/java
Tbh working with c++ is going to be a lot harder then working in python, but who needs sleep anyway
Nobody thinks that, but at the same time, nobody wants to spend time writing a custom kernel, type management, and will simply opt for a python wrapper.
Well I did. To learn, and now, at work, I'm yet to touch c++ for training our models.
I learned x86 Assembly last year, it just blew my mind. It made my C++ code much cleaner and better and I feel I can write more optimized code than I would have without knowledge of Assembly language. Now plan is to bulid my own microprocessor RISC-V softcore on 20K LUTs FPGA and create my own ISA and assembly lanauge and after that make a C complier. Its going to be a fun ride 🤠
I wish I could use x86 assembly, it's really hard to find x64 tutorials. But yeah assembly as a whole is awesome and underrated
@@Smurdy1 there are many online tutorials and courses, books available
I also wanna learn assembly, this comment is my biggest motivation rn
@@denyboi thanks and good luck! take a look at FPGA design as well for embedded systems
@@powerHungryMOSFETany recommendations (sources)?
If you're not a beginner and you want to do machine Learning in c++ do it it going to be better and faster when you understand the stuff and don't want to use any library like me for example. I just want to write my own ml library in cuda and do my own models. If you want to do it... Do it
what if you’re a complete beginner and want to create a personal ai agent
As someone who's proficient in both python and C++... i would prefer C++ any day. It's really interesting to code something new especially for ML projects
You can program rollercoster in to excel or Doom on a thootbrush. Answers the question also pretty well.
I wrote my interpretation of NEAT AI in C++. Wrote a few sample "games" that use this library. One of which is of tank bots that roll around and try to hit a moving target while not hitting friendlies. It works really well and processes much faster than an equivalent Python library. Now I'm writing up a Deep Q Learning library in C++ just because...
It would be wonderful if you made a series of this, trust me
Nice video, I learnt something about Machine Learning from an animated profile picture. 😅😊
You are right about the hardest things for beginners are pointer/references, passing by x, classes and further on templates, smart pointers, function pointers, etc.
But generally, I think C++ is the best language to start with, it is not as complex as people might think/say.
It's like every (normal) programming language, high level languages are of course, extremely easy, but you lack the fundaments of the language by that.
When learning C++, you need to know the fundamentals before you can begin, which I think is really good about the language.
Later on in C++ you will learn about stl, the preprocessor directives, marco's, etc.
So, I dont see any reason why people shouldn't be learning C++ as their first language.
Well, pragmatism would direct people to Python. There is no reason to tech an accountant who uses Excel C. People who are in the science fields don't need complex language features that C++ offers because they will never use them
@@TheBuilder There is no reason for an accountant to learn python.
People in the science fields use applications that doesn't require programming skills.
You'll be surprised.
In India, accountants in some firms need to learn a bit of python to do their work or they spend hours looking manually doing it often on an old PC monitor, not a laptop@@budgetarms. Hmmm why did the reference come here instead of the beginning, anyway, the python they learn is like a 1hr or 3hr crash course on simpler numpy and pandas rather than basics of python
All the ML tools (except a bunch written in Rust) are written in C. So why it wouldn’t be possible to do it in C++???
Scikit-learn is mostly written in Python and Cython (not in C++), and thus it is not interchangeable with any other language. Neither sktime which is also written in Python.
I'm sure there are machine learning libraries written in all popular languages
there are some machine learning libraries built for c++ but its for in house only and not distributed publicly.. they actually use only for production.
Do you have any good learning sources though? I found "Geek For Geek - Machine learning with c++", a book called "Hands on Machine Learning with c++" and the usual library documentaries. But not to much else thats non-Phyton. Are there actually better places to start? Or paths that are usually taken?
(My higher math is ok and c++ is getting there)
try this developers.google.com/machine-learning/crash-course I also enjoyed ruclips.net/video/8SF_h3xF3cE/видео.html
Im using the Eigen library for ML in c++
Interesting, would C++ be actually much faster than python in Machine Learning as well? I heard that overall c++ just kills python in speed, but does it also translate to much faster and power efficient ML? I would like to learn it if so
If you actually want to do machine learning though, your work will probably be using python and the open source you are building off of is probably also written in python
The program I attended put more emphasis on HPC which is why I'm surprised when I see every resource on the net recommending Python.
@@TheBuilderim a machine learning engineer and have never dealt with ML in any language other that python (with the exception of things like custom cuda kernels). I know there are positions that exist that use languages other than python for ML but they are just not as common, so it makes a lot of sense to me that resources would suggest the most commonly used language which also happens to be a very easy language
Is it me or this guy in UNDERATED AF
look at like counts
and views TF?
Good info
Julia is better than anything for ML :?
Yes, C++ good for AI