Try out the course of MIT 18.06 in Linear Algebra from OCW.
If you don’t imagine matrices, then you can ask a sighted person for help. As
soon as you capture the basics, you can count on your understanding of the
topic to understand more advanced operations.
I don’t know a course specific for the blind, but I may make a one if there is
enough students.
Of course, I can help in understanding anything if I can.
Regards,
Mohamed E. Fayed
On 28 Jan 2022, at 8:43 PM, hamidreza abroshan <hamidreza.abroshan@xxxxxxxxx>
wrote:
Thanks all! I'm very excited to hear from geeks like you!
My first problem is that I don't know the required math that you mentioned.
Do you know any accessible math tutorial?
I googled a lot and couldn't find any resources.
The second problem is finding a course or tutorial that teaches this subject
in an accessible way. I tried some courses and could not understand much (I
think because they need a math background.)
Best.
On Fri, Jan 28, 2022 at 8:10 PM Muhammad Fayed <m10fayed@xxxxxxxxx
<mailto:m10fayed@xxxxxxxxx>> wrote:
That’s because all papers’ codes are not in Jupiter notebooks. The codes are
just python files.
According to me in my workflow, I won’t need Colab unless I test something or
some tool in a hurry. But at the end, write script that is callable anywhere
in the system.
Colab is very useful in learning Deep Learning. If you signup for paid
course, they will give you access to resources. You may need to discuss its
accessibility with the course provider or the hosting platform.
Regards,
Mohamed E. Fayed
On 28 Jan 2022, at 6:30 PM, Muhammad Fayed <m10fayed@xxxxxxxxx
<mailto:m10fayed@xxxxxxxxx>> wrote:
BY THE WAY, it is worth mentioning that I won’t need it in my daily work.
In NLP, we edit existing codes like any python project, we don’t require
tracing code much.
If I need something interactive, I just use iPython in terminal and it
works.
Regards,
Mohamed E. Fayed
On 28 Jan 2022, at 6:28 PM, Muhammad Fayed <m10fayed@xxxxxxxxx
<mailto:m10fayed@xxxxxxxxx>> wrote:
That’s good to hear.
What software do you use to access it?
I use Mac OS 12.2 and it is accessible to some extent but not usable to
me. In other words, it will decrease my productivity significantly if I
want to use it.
Regards,
Mohamed E. Fayed
On 28 Jan 2022, at 4:10 PM, <travis@xxxxxxxxxxxxxx
<mailto:travis@xxxxxxxxxxxxxx>> <travis@xxxxxxxxxxxxxx
<mailto:travis@xxxxxxxxxxxxxx>> wrote:
Regarding Jupyter Notebook: I did not understand your comments. The last
I checked, Google Collab was pretty accessible including the code editor.
And the output is very nice e.g., printing a Pandas dataframe gives
actual HTML tables.
That said, for a better editing experience I prefer VS Code's Jupyter
Notebook feature.
-----Original Message-----
From: program-l-bounce@xxxxxxxxxxxxx
<mailto:program-l-bounce@xxxxxxxxxxxxx> <program-l-bounce@xxxxxxxxxxxxx
<mailto:program-l-bounce@xxxxxxxxxxxxx>> On Behalf Of Muhammad Fayed
Sent: Friday, January 28, 2022 6:23 AM
To: program-l@xxxxxxxxxxxxx <mailto:program-l@xxxxxxxxxxxxx>
Subject: [program-l] Re: How much machine learning and deep learning is
accessible for blind users in production?
Hi,
Hope my mail finds you well,
Regarding your question, I’m an associate R&D Engineer in AIC in Neural
Machine Translation team. Our work is mainly in Deep learning. Let me
benefit you.
First, regarding Math, deep learning is not about math only, it is about
data cleaning and how to augment the data to generate larger data and to
benefit the model and make the model robust against small changes. I read
papers a lot of the time, and I just ignore the mathematics since it is
written in PDF and PDFs are known for being poorly accessible in math
content.
You will need to study matrices mathematics, which is annoying at the
beginning, but you can juts know how to do it and map it to single
variables. Matrices equations are not different from single variable
equations to be easily understandable. But, you need to watch out of
matrices dimensions, which is nota big deal when you are used to.
If I need to understand mathematics of a paper, I just request from a
colleague to read it out for me, or search for online explanation if the
paper is influential and not very new.
Second, regarding tools to use, you only need to access a high
performance GPU. That’s the main issue. Personally, I was using Google
Cloud Platform for my final project in the University and access my
virtual machine via SSH protocols. Jupiter Notebooks are not accessible
to the best of my knowledge till now for editing. So, you can’t depend on
Google Colab. Update me if there is any good news about accessibility of
Jupiter Notebooks.
In work, you will work on either MS Azure, AWS, Google Cloud Platform GCP
or your companies cluster and all are accessible via SSH protocol.
So, don’t try to escape from using terminal.
Mostly in your work, you will require dealing with Linux, unless you work
in Microsoft itself. So, you may buy a linux machine for yourself or a
Macbook or whatever Mac device. The terminal across them has slight
changes. I don’t think Windows Subsystem for Linux is mature enough now
to fully function as a linux machine. You should determine which makes
you happier in work.
The life in AI is not so great yet, but it requires time til it becomes
so and may be sighted persons won’t care very soon.
Hope it helps,
Regards,
Mohamed E. Fayed
On 28 Jan 2022, at 1:54 PM, hamidreza abroshan
<hamidreza.abroshan@xxxxxxxxx <mailto:hamidreza.abroshan@xxxxxxxxx>>
wrote:
Hey all,
As I mentioned in my introduction email, I did not study computer
science and do not have a math background.
I am just interested in learning AI and machine learning and want to get
more info about existing tools regarding accessibility.
The other challenge Is math(I think.)
I want to know If a blind dev wants to switch to machine learning and
works in production, What are the most challenges that he should cope
with?
How deep is math knowledge required?
How can he analyze the charts and similar visual graphics?
Do you know any accessible resources?
Thanks for your attention!
--
hamidreza
** To leave the list, click on the immediately-following link:-
** [mailto:program-l-request@xxxxxxxxxxxxx ;
<mailto:program-l-request@xxxxxxxxxxxxx>?subject=unsubscribe]
** If this link doesn't work then send a message to:
** program-l-request@xxxxxxxxxxxxx
<mailto:program-l-request@xxxxxxxxxxxxx>
** and in the Subject line type
** unsubscribe
** For other list commands such as vacation mode, click on the
** immediately-following link:-
** [mailto:program-l-request@xxxxxxxxxxxxx ;
<mailto:program-l-request@xxxxxxxxxxxxx>?subject=faq]
** or send a message, to
** program-l-request@xxxxxxxxxxxxx
<mailto:program-l-request@xxxxxxxxxxxxx> with the Subject:- faq
** To leave the list, click on the immediately-following link:-
** [mailto:program-l-request@xxxxxxxxxxxxx ;
<mailto:program-l-request@xxxxxxxxxxxxx>?subject=unsubscribe]
** If this link doesn't work then send a message to:
** program-l-request@xxxxxxxxxxxxx
<mailto:program-l-request@xxxxxxxxxxxxx>
** and in the Subject line type
** unsubscribe
** For other list commands such as vacation mode, click on the
** immediately-following link:-
** [mailto:program-l-request@xxxxxxxxxxxxx ;
<mailto:program-l-request@xxxxxxxxxxxxx>?subject=faq]
** or send a message, to
** program-l-request@xxxxxxxxxxxxx
<mailto:program-l-request@xxxxxxxxxxxxx> with the Subject:- faq
** To leave the list, click on the immediately-following link:-
** [mailto:program-l-request@xxxxxxxxxxxxx ;
<mailto:program-l-request@xxxxxxxxxxxxx>?subject=unsubscribe]
** If this link doesn't work then send a message to:
** program-l-request@xxxxxxxxxxxxx <mailto:program-l-request@xxxxxxxxxxxxx>
** and in the Subject line type
** unsubscribe
** For other list commands such as vacation mode, click on the
** immediately-following link:-
** [mailto:program-l-request@xxxxxxxxxxxxx ;
<mailto:program-l-request@xxxxxxxxxxxxx>?subject=faq]
** or send a message, to
** program-l-request@xxxxxxxxxxxxx <mailto:program-l-request@xxxxxxxxxxxxx>
with the Subject:- faq
--
hamidreza
