/robowaifu/ - DIY Robot Wives

Advancing robotics to a point where anime catgrill meidos in tiny miniskirts are a reality.

Reports of my death have been greatly overestimiste.

Still trying to get done with some IRL work, but should be able to update some stuff soon.

#WEALWAYSWIN

Max message length: 6144

Drag files to upload or
click here to select them

Maximum 5 files / Maximum size: 20.00 MB

More

(used to delete files and postings)


Open file (14.96 KB 280x280 wfu.jpg)
Beginners guide to AI, ML, DL. Beginner Anon 11/10/2020 (Tue) 07:12:47 No.6560
I already know we have a thread dedicated to books,videos,tutorials etc. But there are a lot of resources there and as a beginner it is pretty confusing to find the correct route to learn ML/DL advanced enough to be able contribute robowaifu project. That is why I thought we would need a thread like this. Assuming that I only have basic programming in python, dedication, love for robowaifus but no maths, no statistics, no physics, no college education how can I get advanced enough to create AI waifus? I need a complete pathway directing me to my aim. I've seen that some of you guys recommended books about reinforcement learning and some general books but can I really learn enough by just reading them? AI is a huge field so it's pretty easy to get lost. What I did so far was to buy a non-english great book about AI, philosophycal discussions of it, general algorithms, problem solving techniques, history of it, limitations, gaming theories... But it's not a technical book. Because of that I also bought a few courses on this website called Udemy. They are about either Machine Learning or Deep Learning. I am hoping to learn basic algorithms through those books but because I don't have maths it is sometimes hard to understand the concept. For example even when learning linear regression, it is easy to use a python library but can't understand how it exactly works because of the lack of Calculus I have. Because of that issue I have hard time understanding algorithms. >>5818 >>6550 Can those anons please help me? Which resources should I use in order to be able to produce robowaifus? If possible, you can even create a list of books/courses I need to follow one by one to be able to achieve that aim of mine. If not, I can send you the resources I got and you can help me to put those in an order. I also need some guide about maths as you can tell. Yesterday after deciding and promising myself that I will give whatever it takes to build robowaifus I bought 3 courses about linear alg, calculus, stats but I'm not really good at them. I am waiting for your answers anons, thanks a lot!
Just letting you know I saw your post OP. I'm one of the two you linked. I'll have to think through before giving you a reply so please be patient. It may take me a few days. The other Anon may be able to reply sooner, as he's much more versed than I am. Good thread topic for /robowaifu/ OP, so thanks for taking the time to spell things out well in your OP. :^)
Open file (184.67 KB 1669x542 freematerials.png)
First of all save your money for hardware and use http://libgen.rs/ for free books, I find the 'handbook' series from the publisher Springer has great work on a variety of topics relating to topics this board discusses. Also you can use https://torrentz2.is/ for online courses, this is what I found doing a search for 'udemy machine learning'. Be careful when pirating as the sites linked to on that torrent search engine aren't always safe. I'd recommend against amassing a horde of learning materials as those courses or books won't do you any good if your goals are unrealistic. Start with the basics moving to more difficult topics and constantly reevaluate your goals.
>>6561 Okay anon, I'll be waiting for your response. Thank you for sparing your time to help me :) >>6566 Oh, thanks for letting me know about that torrent site. Looks like there are a lot of great stuff there. >I'd recommend against amassing a horde of learning materials as those courses or books won't do you any good if your goals are unrealistic Yeah, I figured. That was the reason I bought some basic udemy courses in the first place. Thinking about starting with ML without heavy maths and learn basic consepts such as: Data Preproccessing, Numpy, Pandas, Inferential Stats, Data Visualisation, Linear Regression, Polynomial Regression, Gradient Descent, KNN, Model Performance Metrics, Naive Bayes, Logistic Regression, SVM, SVR, Decision Trees, Decision Tree Regression, Random Forest Regression, Kernel SVM, K means Clustering, Hierarchical Clustering, Apriori, Eclat, UCB, Thompson Sampling, Ensembling, Unsupervised Learning, NLP, PCA, LDA, XGBoost I bought 2 courses which covers the topics above. Both of them claim to have only prerequisite of "some basic high school maths and programming". I guess that those basic highschool math topics are simply Linear Alg, Calculus and maybe a little bit of stats. Course only gives you an idea about how the algorithms works, skipping the heavy math parts and shows you how to implement it using python libraries. I thought about finishing maths and stats courses first and then getting into machine learning but I know that won't happen. I will get bored and never finish those courses if I only focus on finishing them. While studying those 2 ML courses I'll try to learn linear algebra, limits, derivatives, integrals, some trig and stats. And after finishing them I will move on to a more theory-based course which teaches you how to implement those algorithms without using libraries with heavy math. Once I complete them I will move on to deep learning with again programming-based course and then move on to the theory-based course. And then implementing stuff, reading papers, doing kaggle competitions I will get used to create projects. Maybe after that I can start to read some of the books you guys recommended. Is there anything wrong with that approach? Pls give me some feedback.
Open file (114.23 KB 1011x720 __.jpg)
>>6560 >Assuming that I only have basic programming in python, dedication, love for robowaifus but no maths, no statistics, no physics, no college education how can I get advanced enough to create AI waifus? Creating a simple chatbot AI with a seq2seq network in Python is possible without any mathematical knowledge. However, it won't be very good or keep your attention long and you won't be able to improve on it until you understand how it works. At the minimum you should study matrices, calculus, statistics, linear algebra and programming. Once you have a basic understanding of those fundamental topics, then you can jump into learning whatever area of AI that interests you and focus on studying topics that are relevant to you. I'll dump some good textbooks on these later in >>235 For a deeper understanding you should also study trigonometry, discrete mathematics, differential equations, Fourier transforms, and entropy and information theory. In discrete mathematics you'll wanna pay special attention to graphs and search algorithms. I recommend 3Blue1Brown's channel for learning mathematics: https://www.youtube.com/c/3blue1brown/playlists He also did a really good series explaining how neural networks work: https://www.youtube.com/watch?v=aircAruvnKk&list=PLZHQObOWTQDNU6R1_67000Dx_ZCJB-3pi DeepMind has a comprehensive deep learning course here: https://www.youtube.com/watch?v=iOh7QUZGyiU&list=PLqYmG7hTraZCkftCvihsG2eCTH2OyGScc And a good reinforcement learning course: https://www.youtube.com/watch?v=ISk80iLhdfU&list=PLqYmG7hTraZBKeNJ-JE_eyJHZ7XgBoAyb Henry AI Labs explains various AI topics: https://www.youtube.com/channel/UCHB9VepY6kYvZjj0Bgxnpbw Yannic Kilcher covers recent advances in AI: https://www.youtube.com/channel/UCZHmQk67mSJgfCCTn7xBfew Machine Learning Street Talk is a good AI podcast talking about the interesting stuff going on in AI: https://www.youtube.com/channel/UCMLtBahI5DMrt0NPvDSoIRQ Schmidhuber's website covers a lot of interesting topics in AI research: http://people.idsia.ch/~juergen/ >Which resources should I use in order to be able to produce robowaifus? At the minimum you'll need to know 3D modelling, 3D printing, C or C++ programming, electricity, microcontrollers, physics, classical mechanics, servos and power systems. Someone else will have to chime in on this since my specialty is AI. >>6567 All those courses are overkill for a start. You should focus on: >Data Preproccessing, Numpy, Pandas, Inferential Stats, Data Visualisation, Linear Regression, Polynomial Regression, Gradient Descent, KNN, Model Performance Metrics, Naive Bayes, Logistic Regression, SVM, SVR, Decision Trees, Decision Tree Regression, Random Forest Regression, Kernel SVM, K means Clustering, Hierarchical Clustering, Apriori, Eclat, UCB, Thompson Sampling, Ensembling, Unsupervised Learning, NLP, PCA, LDA, XGBoost The other stuff is good to know but more advanced than you really need to know unless you want to study those areas. You should soldier through basic mathematics and statistics first, just enough to be familiar with what you have to study deeper to take on an AI project you're interested in. If you don't understand the mathematical notation (particularly around summation, products, matrices and probability) being used in machine learning courses you'll become completely lost on what to do or learn. Once you dive into machine learning you'll see how important mathematics and statistics are and won't be bored studying them when you can see how much you need to know them. It's a good idea not to learn anything unless you understand why you're learning it, otherwise you may spend years learning stuff you'll never use.
>>6570 > It's a good idea not to learn anything unless you understand why you're learning it, otherwise you may spend years learning stuff you'll never use. This. Not OP, but thanks for the great material in your post Anon. Very helpful.
Open file (26.51 KB 400x400 43044190.jpg)
>>6570 Well, after reading your post I can clearly understand that I'll need maths, physics at the end. >All those courses are overkill for a start. >You should soldier through basic mathematics and statistics first I think I will study both of those 2 courses till the end even if it is overkill. Because rather than only studying maths I would prefer dividing my time fifty-fifty and study both programming/ML and Maths. I don't know which anon you are but one of you said that he isn't good at maths or physics, he improved himself by playing around programming like a lot. I am kinda on the same train but it is pretty clear to me that I will need to learn maths and physics anyways. I believe that if I divide my time in two pieces and study maths and programming/ML at the same time by the end of those 2 ML courses I will have finished Trigonometry, Linear Algebra, Calculus 1-2, Statistics and Probability. After that I can study more advanced maths such as those topics; >discrete mathematics, differential equations, Fourier transforms, and entropy and information theory as well as deep learning. This way I will have time to both improve my math and programming skills. Even though my motivation is to create a robowaifu I want to improve myself as much as I can in the field of AI. I will try to improve my programming too. Starting from today I will start with one of those ML courses and a trig course. I don't think I will be able to get into robotic stuff at least for a year. I will also spare physics for future. Maybe once I get into college for CS I can work with other engineers to learn from them about robotics. Thank you a lot for helping, I wonder what the other anon is going to recommend me. But I think the path I should be following is clear right now. I will take one left step and then one right step so that I won't fall because of lacking skills for maths or programming.
>>6573 OK, Anon, I've made my birthday present to /robowaifu/ today. Thank you as your question was the spur for me to finally dig in and get this ball rolling for us. Hopefully it will be an ongoing help to all of us! Cheers. >>7143
>>6573 Godspeed anon! Remember to have a break if you are getting a headache! Not kind on the old grey matter, some of that stuff.
>>7214 Thank you anon! Sorry for the late reply, I've been busy studying :^) >Remember to have a break if you are getting a headache! I will, thank you. But I wish I hade more brain processing power and ability to focus. I put in ~4 hours daily, I can't focus more than that. I am trying to get it up to 6 hours for now. Any suggestions on that topic?
>>7374 Not that anon, but I'd say just relax and don't unduly pressure yourself. Work hard, but also take time to relax and remember the fun of why you're doing this. Just remember the old adage Rome wasn't built in a day has stood the test of time for a reason. Keep your nutrition levels up and take regular short breaks at the very least.
>>7374 Don't study to hard, this won't make it better. Resting is part of the process. I try to follow some of the advice in A Mind For Numbers: How to Excel at Math and Science, you could also listen to as mp3 (magnet-link in textfile, book and mp3 on limetorrents.info or in the linked textfile, can't upload any of it here). You might listen to the book, while going for a walk, which would have the downside of not being able to make notes, though. https://files.catbox.moe/2bvsmk.epub https://files.catbox.moe/tcm2is.txt
Open file (100.55 KB 638x978 IMG_20201110_233630.jpg)
Open file (936.87 KB 4096x2896 IMG_20201106_032003.jpg)
Open file (77.43 KB 1306x822 IMG_20201104_174127.jpg)
Open file (57.16 KB 840x600 IMG_20201104_145915.jpg)
Open file (94.40 KB 1070x1252 IMG_20201019_070005.jpg)
Stolen from Twitter for us here. Mainly overviews. Cheat sheets will follow.
Open file (102.81 KB 794x902 IMG_20200925_235656.jpg)
Open file (253.39 KB 1573x1019 IMG_20200925_235627.jpg)
Open file (99.12 KB 1418x620 IMG_20200629_050706.jpg)
Open file (211.52 KB 796x668 IMG_20200625_030841.jpg)
Open file (257.29 KB 850x576 IMG_20200625_030831.jpg)
More...
Think I got the last few from here: https://www.datasciencecentral.com/page/search?q=cheat+sheet The last one is an example how to make a list of ressources related to a topic and then learn it step by step, not trying to understand it at the first try.
Thanks for all the effort Anon, much appreciated.
Open file (780.82 KB 2000x3400 ClipboardImage.png)
Open file (540.77 KB 2000x3000 ClipboardImage.png)
Open file (607.05 KB 1250x693 ClipboardImage.png)
Open file (924.91 KB 1224x794 ClipboardImage.png)
Open file (171.42 KB 875x492 ClipboardImage.png)
>>8247 Thanks!
Open file (170.73 KB 875x559 ClipboardImage.png)
Open file (528.55 KB 564x1002 ClipboardImage.png)
Open file (917.49 KB 875x1236 ClipboardImage.png)
Open file (496.18 KB 2200x1700 ClipboardImage.png)
Open file (960.72 KB 2200x1700 ClipboardImage.png)
Open file (1.04 MB 1952x1508 ClipboardImage.png)
Open file (314.05 KB 1024x747 ClipboardImage.png)
Open file (697.77 KB 3300x2550 ClipboardImage.png)
Open file (136.05 KB 1650x1275 ClipboardImage.png)
Open file (575.21 KB 1250x884 ClipboardImage.png)
Open file (287.38 KB 800x566 ClipboardImage.png)
Open file (463.83 KB 1250x883 ClipboardImage.png)
Open file (1.12 MB 1250x883 ClipboardImage.png)
Open file (559.02 KB 1753x1240 ClipboardImage.png)
Open file (482.82 KB 1250x884 ClipboardImage.png)
Open file (1.19 MB 1250x966 ClipboardImage.png)
Open file (1.18 MB 1250x966 ClipboardImage.png)
Open file (1.15 MB 1250x966 ClipboardImage.png)
Open file (1.13 MB 1250x966 ClipboardImage.png)
Open file (579.35 KB 1250x884 ClipboardImage.png)
Open file (491.45 KB 1250x884 ClipboardImage.png)
Open file (1.19 MB 1250x966 ClipboardImage.png)
Open file (1.46 MB 1250x966 ClipboardImage.png)
Open file (1.10 MB 1250x883 ClipboardImage.png)
Open file (285.42 KB 1000x704 ClipboardImage.png)
Open file (347.72 KB 1250x693 ClipboardImage.png)
Open file (327.08 KB 1250x975 ClipboardImage.png)
Open file (261.90 KB 850x1100 ClipboardImage.png)
Open file (165.69 KB 875x555 ClipboardImage.png)
Open file (606.29 KB 1652x1250 ClipboardImage.png)
Okay, thanks. That's the better way to do it.
>>8256 Y/w. There's still the pdfs to get if you're interested.
Here's a good example for a little lecture on how to understand some AI problem better: https://nitter.dark.fail/svpino/status/1357302018428256258#m - I saved it as a pdf, no idea if that's usefull, but that Nitter instance might not be there forever.
>>8447 Neat, thanks Anon.
Open file (81.70 KB 1378x562 IMG_20210208_184145.jpg)
Open file (164.53 KB 594x1937 IMG_20210208_090120.jpg)
Open file (65.89 KB 1276x846 IMG_20210208_072837.jpg)
Open file (238.86 KB 1200x927 IMG_20210208_071849.jpg)
Posted on Twitter by data scientists.
>A curated list of awesome, free machine learning and artificial intelligence courses with video lectures. All courses are available as high-quality video lectures by some of the best AI researchers and teachers on this planet. >Besides the video lectures, I linked course websites with lecture notes, additional readings and assignments. https://github.com/luspr/awesome-ml-courses
>>8495 Thanks! That's a lot of content. It would be great if we can find someplace to keep this stuff safe off of Youtube somewhere.
Has anyone here read the book "Artificial Intelligence: A Modern Approach"? Any thoughts on it?
>>8609 Not yet, but noted in case I haven't already. I guess I downloaded and noted the name of reading material which would take me 10yrs or more to read and work through, though. I'll start with anything beyond simple chatbots, image and voice recognition when I mastered all these basic things, and the stuff I'm already working on, and when have a functional female body which needs to get smarter. Which will be the case in 5-10yrs, I hope. Related: Uploaded file and https://github.com/aimacode
I'll probably go with that, as soon as I have the time: >Dive into Deep Learning: An interactive deep learning book with code, math, and discussions, based on the NumPy interface. >Implemented with NumPy/MXNet, PyTorch, and TensorFlow >Adopted at 175 universities from 40 countries >Each section is an executable Jupyter notebook >You can modify the code and tune hyperparameters to get instant feedback to accumulate practical experiences in deep learning. http://d2l.ai/
Open file (94.73 KB 1362x1104 IMG_20210301_213941.jpg)
Open file (87.07 KB 1267x710 IMG_20210222_233049.jpg)
Open file (137.51 KB 577x385 IMG_20210219_162908.jpg)
Open file (162.50 KB 1588x938 IMG_20210218_113525.jpg)
Open file (100.78 KB 700x1458 IMG_20210313_234930.jpg)
Open file (148.21 KB 1080x1080 IMG_20210313_185721.jpg)
Open file (105.91 KB 577x385 IMG_20210325_191728.jpg)
Open file (64.24 KB 784x427 IMG_20210325_171914.jpg)
Open file (178.76 KB 2334x1025 IMG_20210322_143128.jpg)
Open file (97.96 KB 1207x842 IMG_20210322_084800.jpg)
>>9266 > #1 That actually helps me understand that term a lot better now. I still don't understand the implications of one way or other, but at least I kind of have an idea what it means now. > #2 I know exactly what that is -- personally -- but I can also add that eventually you'll holistically put everything together in your head if you just keep moving forward. Don't quit and you'll succeed. > #3 Somehow, I can understand an idea when it's represented graphically like that, rather than a bunch of 'indecipherable' math symbols of a formula. I don't think I'm alone in this either. > #4 LOL. Surely this is a simple thing to pick up over the weekend. R-right, Anon? :^)
>>9274 Tha last pic is more about having an overview and being aware that there's more than deep learning. In some cases another thing might work better or be good enough but easier to archive. It's also useful to use these terms to look for videos to get an idea about something. Maybe it fits to a problem that one wants to solve.
>>9278 Makes sense, thanks Anon. It's a big field.
Open file (220.57 KB 1199x540 IMG_20210331_191630.jpg)
Open file (52.91 KB 1404x794 IMG_20210331_191334.jpg)
> completely removing the background of a picture (robust PCA) > PCA's main goal: dimensionality reduction. >You can take a bunch of features that describe an object and, using PCA, come up with the list of those that matter the most. >You can then throw away the rest without losing the essence of your object. https://nitter.dark.fail/svpino/status/1377255703933501445
>>9375 That's pretty powerful. I imagine glowniggers are using this idea extensively for surveillance isolation. Not only would this work with a 'pre-prepared' empty background plate for extraction, but a separate system could conceivably create (and keep updated under varying lighting conditions, say) an 'empty' plate from a crowded scene simply by continuously sweeping the scene and finding areas that don't change much frame-to-frame. These blank sections can then all be stitched together to create the base plate to use during main extraction process. Make sense? Ofc, a robowaifu can use this exact same technique for good instead to stay alert to important changes in a scene, alerting her master to anything she sees that might be an impending threat, or even take action herself to intervene. Simplification is the key to both understanding, and to efficiency in visual processing and other areas.
Edited last time by Chobitsu on 05/10/2021 (Mon) 01:01:13.
Related: GPT-2 for beginners >>9371
Illustrated guide to transformers, a step by step introduction: https://youtu.be/4Bdc55j80l8
Edited last time by Chobitsu on 05/10/2021 (Mon) 00:59:53.
Open file (18.21 KB 1600x900 IMG_20210514_143007.jpg)
Open file (24.74 KB 1600x900 IMG_20210514_143019.jpg)
Open file (18.13 KB 1600x900 IMG_20210514_143028.jpg)
Open file (47.79 KB 2400x1350 IMG_20210514_165116.jpg)
Open file (79.86 KB 2400x1350 IMG_20210514_165319.jpg)
This could also fit in the math thread, but it's notation used in ML.
>>6560 Heres a wholistic beginners understanding all that you see, the framework that people take for machine learning is to build a mathematical function that can make a prediction. That mathematical function can be created in many ways, at the end of the day its supposed to provide you with a value of some kind that you action, or output. More recently that function is created using deep learning = parametric system that learns how to capture data and create regions between high dimensional datapoints, to segment partitions of that data to do classification or make predictions through regression, obviously there are many other way to this these are the high level constructions. I would suggest you buy grokking deep learning by Andrew trask, he gives you a really good deep insight into DL, In practice however, a lot of the algorithms we use supplement DL techniques we generally use some older ML algorithms and Feature Engineer through PCA or various other Engineering techniques.
Open file (60.57 KB 700x355 p_value.png)
Open file (6.89 KB 446x291 variance.png)
10 Must-Know Statistical Concepts for Data Scientists: https://www.kdnuggets.com/2021/04/10-statistical-concepts-data-scientists.html
>related crosspost (>>10613, >>10614)
>>10607 Thanks very much Anon! Bookmarked.

Report/Delete/Moderation Forms
Delete
Report

Captcha (required for reports)

no cookies?