Introduce yourself here!

Hi everyone!

This is Ankit from Ann Arbor, Michigan. I am here for my third dose of fast.ai (third time’s a charm!). :slight_smile:

I work as a Controls Engineer at AVL in Plymouth, Michigan, currently working on design, validation and integration of driveline control system software for a customer. Over the past few months, I have worked on a few side-projects including object detection for pedestrians for a bus-manufacturer (LV4 concept; project didn’t finalize but worked on a proof-of-concept) and autonomous vehicle simulation pipeline using AirSim (currently working my way through this).

I’ve always been impressed by fast.ai’s top-down approach and the focus on “learning by doing”. A big thank you to Jeremy and Rachel for making this course possible, and super excited about the fastai development and kudos to the dev community. Hoping to put my last few months’ programming blitz to good use and work on exciting things during v3.

See you all in class!

Cheers! :smile:

5 Likes

Hi everybody. I’m Alberto from Barcelona and am currently working in Germany as software developer. Thank you Jeremy and your team for sharing your knowledge and developing this amazing tool.

1 Like

Hi everybody, I am Prafulla from Mumbai, India and currently working as a Solution Architect. I have done some work in ML and DL. I have been planning to do this course through online videos for a while now, but never got to it. This Live course will help dedicate time-block for every lecture. Thanks Jeremy and team for setting up this live course.

1 Like

Hi Everyone,
My name is Idris Azeez, from Nigeria, a first timer here. Thanks everyone for making this a possibility

1 Like

I’m also working in fraud and fascinated by graph. We have successfully used linking analysis to flag activity, but the fraudsters have adapted. I wouldn’t say supervised learning is ‘useless’ but a lot of patterns do pop up and disappear before the model picks them up. I’ve been working on a cardholder embedding approach to identify first party and third party fraud, and would like to incorporate graph embeddings on the more abstract level.

Hi, I’m Yonatan from Israel. I’m a physicist, currently working in the field of ML. I enjoyed and learned a lot from fast.ai’s previous videos, and am looking forward to this one!

As a side project for this course I would like to design an algorithm that will guess the mathematical function from a graph. Here are some examples…

Can you guess the function of the graph below?

image

and this graph?
image

and this one??
image

I’m sure you have some good guesses! Can an algorithm develop this kind of intuition as well? What is a good way to design such an algorithm? This is what I hope to study as a side project while doing this course. It sounds complex but maybe we can at least create a toy version for a limited amount of functions. You are welcome to join, the data is on me :wink:

7 Likes

hello

it’s great to be a part of fast.ai family

big thanks to @jeremy @rachel @sgugger and fast.ai team for this great learning opportunity

together with fastai fans at cleveland artificial intelligence group we organized study group for DL part 1
we are starting structured data so still time to join us

together with TWIML&AI podcast we organized online ML study group and now diving into random forests
join us if it sounds interesting to you

also at TWIML&AI we have online DL study group and we are excited about the new course
sign up here

Hi. I’m currently an MSECE masters student at Carnegie Mellon University-Africa currently in my first semester. I hope to become a machine learning engineer and I hope to combine this course with fundamentals training for a strong foundation to work in industry or in research.

1 Like

Hi all,

I’m French, working in security evaluation. I discovered the field of machine learning through the Andrew Ng course on coursera 2 years ago, and also the beginning of the DL course. Then someone told me about the fast.ai course, so I gave it a try, and I enjoyed it so far. Fast.ai makes DL very cool for me, and I hope to use DL in several fields.
I started to apply DL to side channel attacks (aiming at recovering secret keys from signal processing) at work, more or less successful. I hope to improve my intuitions and skills with the fast.ai v3 course, and look forward to starting using fastai v1!
Big thanks to all fast.ai team :slight_smile:

2 Likes

Nice, while I am at the Know-Center the main sponsor for my work currently is also AVL, cool. (I work with Adrian Remonda on deep learning topics there [I just started]). Let’s keep in touch!

1 Like

Hi Ralph,
Yeah ‘useless’ is an exaggeration. I have to drum down the hype man :). I haven’t looked deep into third party fraud. I am mostly looking at money laundering and first party fraud (perpetrated by fraud rings). Well technically money laundering is first party fraud from the government’s perspective right ?. I would love to collaborate in any way.

I’m Nate. I live in Boston in the US. I played poker professionally for many years. Last year I quit poker and joined a startup focused on machine learning. I’m also a chess master. Naturally I’m quite interested in the applications of machine learning to games, but also in learning and cognition.

1 Like

Glad to hear that there are some people in the Dallas / Fort Worth (DFW) area that are taking this course. I’m in Richardson (near Dallas) and have been thru part 1 twice so far and understand a bit of it, but I’m back for the third time. I wonder if there are enough people here in the DFW area to form a study group. If anyone else is in DFW, please let me know.

Pradeep,

I do think graph has a lot more potential with first party and AML. Third party is somewhat random. It might have a node or two in common - useful for tracking activity but not something you could use for pattern identification in new data.

Definitely interested in collaborating. We should be building a graph db on a new server in the next couple of weeks so a bit of a sandbox to play in. I’d love to see what we can cook up.

My name is Bryan Smith. Currently, I am an Applied Machine Learning Engineer on the Commercial Software Engineering team at Microsoft, where I developed cutting-edge ML solutions with customers. Previously, I wrote behavioral detections to find malicious activity in Azure service logs for the Azure Security team. Outside of work, I have been using machine learning to fill out NCAA tournament brackets. My models predicted Butler’s run to the Final Four in 2010 and Dayton’s run to the Elite Eight in 2014. I also got 12th place in the AzureML March Madness 2015 competition and got 5th place out of 40 in the AzureML March Madness 2016 competition.

I hope to get a better understanding of ULMFit in order to do named entity recognition. I am also interested in doing activity detection in video.

3 Likes

Welcome Bryan - I’m Brian Smith from Microsoft too. ML stuff isn’t my day job but looking to learn.

Hi, I’m Alvaro. My current day job is Data Engineer. Interested in cognition and physics since childhood. A beginner for Machine / Deep Learning, I have sit thru some presentations but no in-form courses yet. My hands-on experience is limited to programming some weeks ago a simple neural network for MNIST to get more familiar with NN concepts and with Python + Numpy and then PyTorch + GPU, for a not so good 98.19% success rate in a laptop.

I went to read more to decide where to go next, and then found out about fast.ai and all the amazing stuff being done here. I see this particular fast.ai course as the best way to jump into the journey from spectator to practitioner.

As I have not done any ML/DL project yet, I may be off-base on these projects below:

  • For my day job, I will be looking for a way to apply what I learn on deep learning in this course to SQL Optimization.

  • A possible long term side project, no idea how non-sense it is or if it’s already done.
    Emotions. Can we use deep learning to identify human emotions in text, handwriting, photo, voice, or video? Can we differentiate emotions? If so, how many? Can we use deep learning to confirm if they are 2, 4, 6, 8, 9, or 27 as different researchers say? Is the emotion map, in positions or intensities, the same from person to person? Do individuals have emotional fingerprints? Can some of the emotions be expressing at the same time? Are there velocities in their change? Can we predict their changes, even if for a very short time? Can we find the same set of emotions in other animals? Can we find a set of emotions in the behavior of trained models? More importantly, can we get something out of this to improve lives?

  • I may as well join another project that makes more sense.

fast.ai is helping make this world better. More hands may assist in spreading its good influence sooner and wider.

That make 3 of us, send you PM bro, please check, waiting for other Indonesian (if any) comes up :smiley:

Hi @saputro & @rifqiabidin. With you guys now we got 5 Indonesian, haha. Please send your whatsapp contact PM to me, perhaps we could form group chat to discuss Jeremy class.

Hi, I’m in Shanghai China. Currently working on a career switch from traditional industry to AI. Similar to quite a few other folks, I’ve went through Andrew Ng’s ML course and DL specialization, and then found FastAI to be a great place to further advance and get practical. And to my great surprise, instead of just get to know what’s the “norm”, I actually get to be on the cutting-edge! WoW!
Currently, I’m in the middle of Part1v2 and are really enjoying the course. Look forward to the live v3.
Big thanks to the FastAi team for offering such a fantastic course.

2 Likes