New work yao qin, Nicholas Carlini, ian good fellow, and Gary Cottrell on generating imperceptible, robust, and targeted adversarial examples for speech recognition systems! Paper: https://arxiv.org/abs/1903.10346 Audio samples: http://cseweb.ucsd.edu/~yaq007/imperceptible-robust-adv.html

✴️ @AI_Python_EN

Knowledge Graphs: The Third Era of Computing

Blog by Dan McCreary: https://lnkd.in/epZU-Yi

#MachineLearning #KnowledgeGraphs #AI #ProceduralProgramming

✴️ @AI_Python_EN

Francesco Cardinale

I'm happy to announce that we just open-sourced a major update for our image super-resolution project: using an adversarial network and convolutional feature maps for the loss, we got some interesting results in terms realism and noise cancellation.
Pre-trained weights and GANs training code are available on GitHub!
If you want to read up about the process, check out the blog post.
Also, we released a pip package, 'ISR' (admittedly not the most creative name:D), with a nice documentation and colab notebooks to play around and experiment yourself on FREE GPU(#mindblown). Thanks to Dat Tran for the big help.

💻Blog: https://lnkd.in/dUnvXQZ
📝Documentation: https://lnkd.in/dAuu2Dk
🔤Github: https://lnkd.in/dmtV2ht
📕Colab (prediction): https://lnkd.in/dThVb_p
📘Colab (training): https://lnkd.in/diPTgWj

https://lnkd.in/dVBaKv4

#opensource #deeplearning #gans #machinelearning #keras

✴️ @AI_Python_EN

Abstract Art with ML: Com­po­si­tion­al Pat­tern Pro­duc­ing Networks

Randomly initialised neural networks are able to produce visual appealing images. By Jan Hünermann: https://lnkd.in/dKQEpE6

#art #artificialintelligence #machinelearning

✴️ @AI_Python_EN

**** CS294-158 Deep Unsupervised Learning Spring 2019 ****

1) Week 1
https://lnkd.in/f8RbXpT

2) Week 2
https://lnkd.in/fig-KwB

3) Week 3
https://lnkd.in/fQkiQVh

4) Week 4
https://lnkd.in/f-euBDk

5) Week 5
https://lnkd.in/fz-Aqq2

6) Week 6
https://lnkd.in/fZpSG-Y

✴️ @AI_Python_EN

What about with JSON? 😂😂

✴️ @AI_Python_EN

Teaching deep generative modelling techniques for NLP applications at Yandex NLP week https://academy.yandex.ru/events/data_analysis/NLP_week/ … All material available from https://github.com/philschulz/VITutorial/tree/yandex2019/modules

The US Military Is Creating the Future of Employee Monitoring

"A new AI-enabled pilot project aims to sense “micro changes” in the behavior of people with top-secret clearances. If it works, it could be the future of corporate HR."

In Defense One: https://lnkd.in/e6iuJU6

#ArtificialIntelligence #DeepLearning #MachineLearning

✴️ @AI_Python_EN

Articles on Natural Language Processing (NLP)
by Vincent Granville
https://lnkd.in/fnBGQfM #technology #naturallanguageprocessing #artificialintelligence #businessintelligence #deeplearning #machinelearning #python #datascience

✴️ @AI_Python_EN

Word2vec was great .... in 2013. There's a lot of value in the idea of embeddings still (although, for a lot of NLP tasks, you should really upgrade to deeper representations. Check out transformer networks from 2017 and what followed them).

In any case, this is a modern analysis that should make clear what the possibilities of embeddings created with word2vec and similar techniques are.

Do pay attention to the genealogy of techniques that followed word2vec, as the original idea is powerful.

https://lnkd.in/f5KBf76

#neuralnetworks #artificialintelligence #machinelearning #ai

✴️ @AI_Python_EN

#AI hype can certainly be irritating, and many AI researchers have expressed concerns that it may lead to yet another AI winter.

At the opposite extreme are apocalyptic visions of a future in which "the machines" enslave or even eliminate humanity. Eschatology in its latest incarnation.

In keeping with ancient statistical traditions, I personally am wishy-washy about AI and technology in general. AI represents both an opportunity and a threat, but I don't know what the future holds in store for me or anyone else.

I'm not counseling indifference, though. As with any issue we regard as important, I think we should stay calm and do our homework.

✴️ @AI_Python_EN

Understanding the need to use mean, median and mode.

Understanding the need to use Inter-quartile ranges and not normal ranges.

Understanding the use of a line chart instead of a bar chart.

Understanding the model parameters and stats behind it to create a better model.

All these questions and many more can be answered, if we have a stronghold of the basic/intermediate level of statistics.

Even though I have a decent understanding of statistical concepts and always rely on reading up the stats part of models when I'm using them. I picked up head first statistics book this week.

After a long time, I felt I had spent a good 2 hours of the day to learn something new apart from my daily business analytics work.

I'll summarize my learning in a blog post or article after completing the book and also suggest this book to anyone who wants to learn statistics for doing data science.

✴️ @AI_Python_EN

Pranav Dar in his Medium post has classified the pretrained models into three different categories based on their application and here it is!

Multi-Purpose #NLP Models: ULMFiT, Transformer, Google’s BERT, Transformer-XL, OpenAI’s GPT-2.
Word Embeddings: ELMo, Flair.
Other Pretrained Models: StanfordNLP.

🌎 Link Review

✴️ @AI_Python_EN

Yoshua Bengio, Geoffrey Hinton and Yann LeCun, the fathers of #DeepLearning, receive the 2018 #ACMTuringAward for conceptual and engineering breakthroughs that have made deep neural networks a critical component of computing today. http://bit.ly/2HVJtdV

✴️ @AI_Python_EN

Tracking Progress in Natural Language Processing

By Sebastian Ruder: https://lnkd.in/e6tkGHH

#deeplearning #machinelearning #naturallanguageprocessing

✴️ @AI_Python_EN

One of my favorite tricks is adding a constant to each of the independent variables in a regression so as to shift the intercept. Of course just shifting the data will not change R-squared, slopes, F-scores, P-values, etc., so why do it?

Because just about any software package capable of doing regression, even Excel, can give you standard errors and confidence intervals for the Intercept, but it is much harder to get most packages to give you standard errors and confidence intervals around the predicted value of the dependent variable for OTHER combinations of the independent variables. Shifting the intercept is an easy way to get confidence intervals for arbitrary combinations of the independent variables.

This sort of thing becomes especially important at a time when the Statistics community is loudly calling for a move away from P-values. Instead it is recommended that researchers give confidence intervals in clinically meaningful terms.
#data #researchers #statistics #r #excel #regression

✴️ @AI_Python_EN

SciBERT: Pretrained Contextualized Embeddings for Scientific Text

Beltagy et al.: https://lnkd.in/eAT3mSK

#ArtificialIntelligence #DeepLearning #MachineLearning

✴️ @AI_Python_EN

All Data Science ***Cheat Sheets*** in one place.


Github link - https://lnkd.in/fGeGXQs

#datascience #machinelearning #excel #deeplearning #python #R

✴️ @AI_Python_EN

GAN Lab: Play with Generative Adversarial Networks (GANs) in your browser!

https://lnkd.in/dfiFvrc

Research paper: https://lnkd.in/eeYFK4J

#AI #ArtificialIntelligence #GenerativeDesign #GenerativeAdversarialNetworks

✴️ @AI_Python_EN

How about you?
✴️ @AI_Python_EN

The Illustrated BERT, ELMo, and co. (How NLP Cracked Transfer Learning)

✴️ @AI_Python_EN