Data science jobs are going to be in increasingly massive demand. Prepare your resume!

Internet of Things companies will dominate the 2020s.

https://www.androidauthority.com/internet-of-things-companies-1023404/

“Each year, Internet of Things companies tell us “this is the year of IoT,” and each year, we get an expensive fridge that no one buys. But IoT is coming. In fact, it is already here! There are currently 6.7 billion “data collecting devices” in use today, with 20 billion projected for 2020 according to Amazon. Gartner predicts there will be 25 billion connected devices by 2021, and Accenture suggests that the global IoT market will be worth $14.2 trillion in 2030 (as reported by CRN).”

“Data science jobs are going to be in increasingly massive demand for these companies. Collecting all that data from users is only useful if businesses know what to do with it, and how to infer actionable advice from it. How do you turn billions of purchases across millions of users in multiple different countries into a better marketing campaign? That’s where data science comes in.”

"One of the biggest areas of concern for Internet of Things companies is data security."

Our conceptual understanding of how best to represent words and sentences in a way that best captures underlying meanings and relationships is rapidly evolving. Among the latest milestones are BERT and ELMo.

This article talks about the concepts one needs to be aware of to properly get his/her head around BERT.

http://jalammar.github.io/illustrated-bert/

#BERT #EMLO #NLP

Transformers and self-attention explained from scratch, also in pythonic language.

#BERT #GPT_2 #Transformer
#python

http://www.peterbloem.nl/blog/transformers

If computers are so smart, how come they can’t read?

https://www.wired.com/story/adaptation-if-computers-are-so-smart-how-come-they-cant-read/

Do you also see the AI’s reproducibility crisis? Is it also the concern that you have? Is it partly due to the black box ML systems?

https://www.wired.com/story/artificial-intelligence-confronts-reproducibility-crisis

Interesting simulations on epidemic spread and effect of social distancing. Takes about 25 min

https://youtu.be/gxAaO2rsdIs

#DataScience #socialdistancing #simulation

Anyone experienced this yet 😐 #covid19 #workfromhome

Hi Folks,

Many of you probably follow sentdex (Harrison Kinsley) on YouTube (channel link below). He has great introductory content for #machinelearning.

Recently he also wrapped up a Kickstarter campaign to write a book on how to implement #deeplearning from scratch in #python. The videos for the book are free (on his channel) and a link to the book's site can be found below.

sentdex youtube -> https://lnkd.in/gXkhEv7

kickstarter -> https://lnkd.in/gcVyRa6

book site -> https://nnfs.io/

10 times less energy to train deep neural networks with Early Bird - Towards making AI both more environmentally friendly and more inclusive.

#DNN #AI #ML

https://www.sciencedaily.com/releases/2020/05/200518144908.htm

Reducing downtime with predictive maintenance in Shell Pernis.

https://amp-ft-com.cdn.ampproject.org/c/s/amp.ft.com/content/bfbac636-ee8b-11e9-a55a-30afa498db1b

Python at work 🐍 #python

Special Report

Future of AI and Digital Healthcare


https://www.ft.com/reports/future-ai-digital-healthcare

#AI #ml

Microsoft has built one of the top five publicly disclosed supercomputers in the world, making new infrastructure available in Azure to train extremely large artificial intelligence models. #MSBuild2020 #developer #dev #AI #ml

https://blogs.microsoft.com/ai/openai-azure-supercomputer/?_lrsc=e416cecb-3272-4988-a27c-f3b786153103

AI failures aren’t due to AI models: they fail in data prep and organization. Successful AI models require collection and organization of data first.

https://www.ibm.com/downloads/cas/W16VXMEN

A very interesting video about manipulating the RNG (Random Number Generator) in Zelda Windwaker to remove chance in speedruns - by using Bayesian Statistics. By creating prior beliefs about where they are in the RNG state and updating these beliefs based on additional observations, they solve the battleship minigame without leaving anything to chance. This might be one of my favorite Youtube videos and shows how you can combine statistics, domain knowledge and creativity into a solution to a seemingly unsolvable problem.

Highly recommend watching this!

#statistics #gaming #RNG #bayesian

https://m.youtube.com/watch?v=1hs451PfFzQ

With great power comes great responsibility. And this statement fits exactly the current applications of #artificialintelligence, like #machinevision #facialregonition.
There are two main concerns here - one is #privacy, the other one is the cost of an error or misclassification.
While most of us are busy in developing algorithms and applications, it's important to keep these concerns in mind.
And then only we can move from #ai to responsible AI #responsibleai.

This latest one from Last week tonight is an eye-opener.


https://m.youtube.com/watch?v=jZjmlJPJgug&feature=share

Explainable quadratic function #it #explainableAI

There is a new kid on the block for serving #artificialintelligence models. FastAPI is quickly replacing #flask as the go to #python framework. It’s simple to use and... fast, great combination. Checkout the docs and repo

docs -> https://lnkd.in/eaqxxdS
repo -> https://lnkd.in/ewSVm-a