Yes it’s totally fine, but note that we do assume some knowledge of a couple of things that you won’t have done by year 10:
Derivatives / gradients
Logarithms / exponentials
Also, depending on your school, you may not have done matrix multiplication yet.
These are all topics, however, that Khan covers very well. Many adults have forgotten these topics by the times they get around to doing our course - so when I introduce them I mention at the time that it’s something people should check out on Khan if they’re not comfortable with the topic.
Also, I wouldn’t suggest using autonomous vehicles as a 1st project - maybe start by trying to classify individual video frames as “pedestrian”, “car”, “intersection”, etc?.. Once you’ve finished the book, you should be in a position where you can tackle video data.
I’m Fahime. I’m M.Sc in Medical physics and I have no background in programming and deep learning. I’m so intetested in application of deep learning in medical imaging, radiotherapy and so on. Despite that deep learning resourses and facilities are rare in my country, I’m trying to learn it hard. Many times I’m confused because I’m a beginner in deep learning but I hope one day I learn it.
So happy to be here with you
I will follow the invitation to introduce myself. My background is in biomedical engineering and healthcare and rehabilitation technologies where I finished my masters now almost 10 years ago. I have started my career in marketing for a global industry leading company but moved to software development quickly, which is my true passion. My developer experience is in Java, C++, Python (let us not mention C#) where I have worked on projects involving image processing, automation, 3D printing, radiology, dicom, and even AI (however I was not involved in the training or data mining then). In my current project I am designing a UI for an embedded medical device which does not involve deep learning. While I am still curious about deep learning, I chose this course to learn. Thanks to the fast.ai framework, YouTube, paperspace and the kindle ebook I can literally learn about and even do deep learning on my phone whenever there is time. For free!
Hello, I’m from Saint-Petersburg, Russia and my background is video game industry.
Recently I realized that I’d like to work more with interdisciplinary fields, and things that affect real world more directly than videogames do. ML seems like a thing in that direction with many fun properties.
This is my third dive into ML and I hope with fastai I’ll start using it “for real”.
Make something cool y’all! Cheers.
First off, thanks for offering this wonderful course and for bringing the power of deep learning to everyone. My name is Ossian, and I’m from Sweden but nowadays I live in California. My background is in computational physics. I love to build numerical codes for scientific applications like simulating how waves propagate throughout the earth during an earthquake. I’m interested in learning how AI can accelerate scientific applications far beyond what we can achieve with classical approaches.
I’m used to approaching these problems by starting from a well-known mathematical model, i.e. partial differential equations (PDEs), like the wave equation, that describes the physical phenomena. While the PDEs are often quite simple to state (you can write them down in much less than a sheet of paper), finding the actual solution is impossible without seeking a numerical approximation. Once you start thinking about how to solve these types of problems using computing, that’s when the challenges begin. There are complexities at all scales imaginable, in both space and time. From geometric complexity to interactions between different physical systems, and so forth. The way I currently approach these problems is via a classical numerical approach. That is, you discretize space by filling it up with points that represent the locations at which you want to obtain the solution. At each of these points, you approximate the PDE. That involves approximating derivatives and that typically requires forming connections between the points in either an unstructured or structured manner. You run into all sorts of interesting issues like making sure that your solution is well-resolved enough for your application and that your method doesn’t cause it to blow up, which can easily happen. Also, depending on the application and numerical approach taken, the resulting computation can become large enough to consume hundreds to thousands of GPUs.
I’m somewhat new to machine learning and deep learning in general, but I’m watching with excitement as these approaches make their way into computational physics. But I don’t want to stand idly by as the developments unfold, I want to be part of it. So, I’m taking my first steps into this exciting field. Today, I joined the deep learning study group, and it is thanks to them that I’m writing this introduction :). If there’s anyone else interested in AI for computational physics, I hope to connect with you. It is always fun to meet new people who share similar interests. If you are interested in following me on my adventure, my Twitter is @ossian_oreilly.
I am a radiologist who inspired so much from Jeremy s last youtube lecture at freecodecamp channel. I suppose radiologists have to be a good model trainer and error analyzer in near future.
After watching video of J.Howard, I have learned that having lots of modified downsampled training dataset is the first step for training an effective model.
So I wrote a code in python for practical bulk converting and downsampling your dicom images to jpeg files. You can try it from the link below DICOMtoJPEGBulkConverter
First of all, thanks for this great course! I really appreciate such a high quality course being available for everyone (for free) online. Keep up the good work!
My name is Aksel and I’m from Amsterdam. Over a decade ago I got a BSc in Artificial Intelligence from the University of Amsterdam. After a few career sways I’m now back to programming again. I recently followed a course on product management for AI which triggered my interest in machine learning again.
I’ve started the course a few weeks ago and I’m now in chapter 6, rebuilding the bear classifier for multi-labels. I really enjoy the hands-on approach to building tangible projects.
Looking forward to building many more interesting applications with FastAI.
Hi everyone, i just got my bachelor’s degree in computer science, gonna pursue a graduate degree now. Just finished the fifth chapter of the book. I’am really motivated and looking forward to help/ contribute/ participate in computer vision related projects .
I’m Paolo from Italy and I’d like to thank you fast.ai and Jeremy for the great knowledge he’s sharing with us.
I’m a graduate in media science and I’m really fascinated about deep learning. I started learning python and data science about one year ago and took some course about ML, accidentally I come up here after googling information on validation sets .
Well, my goal is to gain knowledge and develop a carrier in Data Science and DL, although I don’t have a specific scientific background.
I graduated with Electrical Engineering and worked as a Hardware Validation/Test Engineer for past 10+ years and looking for a career switch into Deep Learning/ Computer Vision. I went through fastai part 1 (2018) version and currently going through the 2020 version. My goal is to get my foot in the door without having taking the traditional university route. If anyone has any tips, it would be greatly appreciated.
My formal education is in physics, I have been a private tutor in math/physics/music for the past 20 years. I’ve been feeling the need for a career shake up for a while now and Covid/lockdowns has been a good catalyst for that. Thought learning Python was a good start and ended up here, I can’t contain my excitement!
Interests include poker, playing flute, psychology, nature aquariums, nutrition. I look forward to being able to apply ML to some of these.
A big thank you to Jeremy et al, you have provided me with a solid path to follow to get started.
My name is David Guevremont. I have a background in Mechanical Engineering and I am from Montreal, Canada and now lives in Knoxville, TN. I Worked for 12 years in fraud detection for a electric utility company. I project managed hardware and software development to detect and stop energy theft on the distribution network. Also, did a lot of Building Energy Modeling.
Bored with that job, I quit and started a food truck in Knoxville TN. I ran it for 3 years. After that chapter closed , I moved to Baton Rouge LA for 3 years where I stayed home to take care of my newborn son while my wife was pursuing her career.
During that time I reflected on my next project…In 2017, I started to take python coding classes online at night. During that time I stumbled on Fastai and started to watch the MOOC in 2018. My interest was peaked…and tried to learn it but with limited time on my hand it proved very difficult to spend enough time on it to master it.
I recently decided started a new job as a python programmer for a medical software company and I am thinking about switching to AI or Data science. I just don’t know how to make that switch yet…
I would be glad to exchange with people about all kinds of topics. I am a very curious person with various experiences…
So far no real project realized with fastai
Nice to meet you all
My twitter handle David@zestybain if anyone wants to reach out
I am a researcher in behavioral science ( focus on behavioral economics and consumer psychology). I have got my Ph.D. from the University of Waterloo. In behavioral science, researchers are curious to find what is happening in people’s minds while making decisions in different contexts, using psychology and experimental studies, and I am incredibly interested in applying the tools such as machine learning models in this matter. I am currently working in an innovation lab of a big payment company in Canada as a postdoc researcher and very welcome to any collaborative research and work on any topics related to behavioral science.
Here is a post I wrote on Medium after I was encouraged by @jeremy in the first version of the Fastai class (introduction to machine learning).
Thanks for making fast.ai available, @jeremy. The Practisioner community owes a lot.
I’m Rajavel KS from Chennai, India.
I had always been a person who was enthusiastic about solving problems. I developed short vision due to facing computer screens and decided before starting my college that I won’t get into a position which will never use computers. Little did I knew after getting my practical exposure to the world that computers are always going to be part of problem solving in one way or the other. While, I had done my Undergrad in Mechanical Engineering, hoping to become an engineer without facing computers, I flunked in many subjects due to lack of interest.
I then did my MBA wihout much focus and started my career in Investment banking.
Now, currently working my way up to have a career in a computational field, irrespective of domain.