How do I get a job in machine learning?

I get asked this question almost daily.

The short answer is "there's no one set path."

But this could fit with anything. And it isn't really helpful.

Here's what I would do if I was fired tomorrow.

💡1. Get really good at whatever it is I decide to do (I'm working on this anyway)

Whatever field you want to get into, #machinelearning, #datascience, health, media, this is a given. You have to be good at what you do.

Not the best at one thing?

That's okay. Combine it with something else and become the best at the crossover.

❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

For me, my #coding skills were lacking, so I made it up for it through several different forms of communication (see point 2).

🎤2. Communicate whatever it is I'm good at in a way people can understand

If you can #code but can't write practice writing.

I can't count the number of times I've found some brilliant code online but struggled to understand it because it hadn't been documented well.

Got a project you've worked on? (you should)

Share it. And share your thought process around each step. Why did you do that thing in part 3?
It's the Winston Churchill approach. He wasn't anywhere near the most qualified person for the role of Prime Minister. But he was the best at communicating what he knew. Because of this, the people put their trust in him. Then he backed it up by being good at what he did.

📝3. Apply relentlessly

If I wasn't getting rejected once per week, I'd start applying more. And not just with a resumé.
Someone told me their resumé gets filtered. I'm not sure what a resumé filter is but I'd find a way to get around it.
Plus, I'm not interested in somewhere that hires solely off the basis of an A4 sheet of paper.
I'd go out of my way to find who the best person would be to talk to. And then I'd find a problem they're having and fix it.
Will this work?
There's no guarantees. There's never a guarantee.
But what's the alternative?
To not be good at what you do?
Or to not communicate your skillset?
Or to not find the right person to talk to?
You already know the answer to these.

❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

The Neural Aesthetic

Notes and around 30 hours of video lectures are up here, by Gene Kogan: https://lnkd.in/dCMpmGx

Big map of all the slides: https://lnkd.in/dB8yJtp

#artificialintelligence #deeplearning #generativemodel #machinelearning #reinforcementlearning


❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

"A Brief Survey of Deep Reinforcement Learning"

Arulkumaran et al.: https://lnkd.in/dwt_--X

#100DaysOfMLCode #ArtificialIntelligence #DeepLearning #MachineLearning #NeuralNetworks #ReinforcementLearning

❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

2018 Year-in-Review: Machine Learning Open Source Projects & Frameworks

https://heartbeat.fritz.ai/2018-year-in-review-machine-learning-open-source-projects-frameworks-430df2fe18cd

#ArtificialIntelligence #DeepLearning #MachineLearning #NeuralNetworks #ReinforcementLearning

❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

An interesting collection of surprising snippets and lesser-known Python features.

>>> WTF() == WTF() # two different instances can't be equal
False
>>> WTF() is WTF() # identities are also different
False
>>> hash(WTF()) == hash(WTF()) # hashes _should_ be different as well
True
>>> id(WTF()) == id(WTF())
True

❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

An interesting collection of surprising snippets and lesser-known Python features.

pip3 install wtfpython

class WTF(object):
  def __init__(self): print("I")
  def __del__(self): print("D")

Output

>>> WTF() is WTF()
I
I
D
D
False
>>> id(WTF()) == id(WTF())
I
D
I
D
True

❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

Professor Andrew Ng:

I’m excited to share with you the AI Transformation Playbook: https://lnkd.in/gJMf_Pq

Drawn from my experience leading Google Brain, Baidu’s AI Group, and Landing AI, this Playbook provides a roadmap for your company to transform into a great AI company.

You can download a free copy of the AI Transformation Playbook here: https://lnkd.in/gJMf_Pq

Building strong AI capabilities for your company will require a long-term investment, but it is feasible for most enterprises. The AI Transformation Playbook will walk you through the following five steps:

1. Execute pilot projects to gain momentum
2. Build an in-house AI team
3. Provide broad AI training
4. Develop an AI strategy
5. Develop internal and external communications

Many of the biggest untapped opportunities in AI lie outside the software industry. I hope that this AI Transformation Playbook will help your company become an AI leader in your industry vertical.

Download your free copy of the AI Transformation Playbook here: https://lnkd.in/gJMf_Pq


❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

Professor Yann LeCun
‏
At NYU, even the art school has a course on deep learning, complete with ConvNets, GANs and RL for artistic creation.

The class is called "neural aesthetics" and is taught by Gene Kogan in the ITP Master's program at the NYU Tisch..

http://ml4a.github.io/classes/itp-F18/

❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

Understanding Supervised, Unsupervised, and #ReinforcementLearning (RL) — getting under the hood with RL: http://bit.ly/2BNrJ1u #abdsc #BigData #DataScience #MachineLearning #Algorithms #AI



❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

PyText: Industrial-strength open source NLP package from Facebook AI: develop NLP models on PyTorch and deploy through ONNX.

Pre-trained models for text classification, sequence tagging, joint intent-slot...

https://t.co/phn8mu9Dhz

❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

Best paper award #NeurIPS2018 main idea: Defining a deep residual network as a continuously evolving system & instead of updating the hidden units layer by layer, define their derivative with respect to depth instead. Paper: https://arxiv.org/pdf/1806.07366.pdf … Code: https://github.com/rtqichen/torchdiffeq

❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

✨📒 Did you know that you can connect to GoogleColab using a local runtime, or a virtual machine running in the cloud (AWSCloud, GoogleCloud, Azure, etc.)? 👉 Check out our guide + blogpost for how to set up your environment: https://research.google.com/colaboratory/local-runtimes.html … https://blog.kovalevskyi.com/gce-deeplearning-images-as-a-backend-for-google-colaboratory-bc4903d24947


❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

Stanford Tracking Artificial Intelligence Research To See Future - Palo Alto, CA Patch

Read more here: https://ift.tt/2PCc1JY

#ArtificialIntelligence #AI #DataScience #MachineLearning #BigData #DeepLearning #NLP #Robots #IoT


❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

linux
❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

What's the hardest part of ML? The most expensive? The most time-consuming? Choosing from:
- data collection & labelling
- data cleaning
- modelling / science
- implementation
- infrastructure / cloud SysOps
- deployment
- maintenance

❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

Attention Networks with Keras The "Attention Network" is one of the most interesting advancements in natural language processing. So, what makes an attention network tick & why it's special?

https://buff.ly/2LNaK0K

#NLP #NeuralNetworks #AI

❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

hardest: features and parameters of the model, most expensive: data collection, cleaning and labeling, most time consuming: multiple iterations in order to converge to the optimal parameters, testing & evaluation.

Dr François Chollet

This is a great answer and I agree -- modelling/science is the hardest (if you want to do it right), and also the most time-consuming due to lengthy iterations. Meanwhile data collection and labelling is the most expensive, and often the most important to the success of a project.

❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

"A Brief Introduction to Machine Learning for Engineers"

By Osvaldo Simeone: https://lnkd.in/eT9FVYd

#ArtificialIntelligence #MachineLearning #NeuralNetworks


❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN

A Full Hardware Guide to Deep Learning

By Tim Dettmers: https://lnkd.in/emiGW6p

#ai #deeplearning #gpu #gpus #hardware


❇️ @AI_Python
🗣 @AI_Python_Arxiv
✴️ @AI_Python_EN