Ladies, if he:

- requires lots of supervision
- yet always wants more power
- can't explain decisions
- optimizes for the average outcome
- dismisses problems as edge cases
- forgets things catastrophically

He's not your man, he's a deep neural network. #AIFun

✴️ @AI_Python_EN

Interesting work from Ross Wightman comparing something like EfficientNet / ResNet which uses only Imagenet data to the Facebook-IG ResNext that was trained on a lot of instagram public data. While their validation scores are close, the test scores seem to diverge more.

FacebookAI ResNeXt models pre-trained on Instagram hashtags stand out in their ability to generalized to the 'ImageNetV2' test set.
#PyTorch
https://colab.research.google.com/github/rwightman/pytorch-image-models/blob/master/notebooks/GeneralizationToImageNetV2.ipynb

✴️ @AI_Python_EN

Towards Declarative Visual Reasoning… Or Not? https://buff.ly/2UkT3u4
#NeuralNetworks #DeepLearning #AI

Neural Architecture Search at CVPR 2019

https://drsleep.github.io/NAS-at-CVPR-2019/

✴️ @AI_Python_EN

Anyone working with data should be concerned about missing data...or they shouldn't be working with data.

One way to address missing data is multiple imputation. This is a complex topic, and many textbooks and articles have been published about it. Four I can recommend are:

- Applied Missing Data Analysis (Enders)
- Handbook of Missing Data Methodology (Molenberghs et al.)
- Flexible Imputation of Missing Data (van Buuren)
- Outlier Analysis (Aggarwal)

The Enders book is probably the most basic and readable of the ones I've listed.

The Stata 16 reference manual on multiple imputation, linked below, also provides a good overview of this subject, and includes many examples and a glossary of key terms.

✴️ @AI_Python_EN

An Awesome Unified Framework of Online Learning Algorithms for Training Recurrent Neural Networks.

Authors here present a framework for compactly summarizing many recent results in efficient and/or biologically plausible online training of recurrent neural networks (RNN).

Their framework organizes algorithms according to several criteria:
(a) past vs. future facing,
(b) tensor structure,
(c) stochastic vs. deterministic, and
(d) closed form vs. numerical.

These axes reveal latent conceptual connections among several recent advances in online learning.

Paper: https://lnkd.in/dTcMbyK

✴️ @AI_Python_EN

Deep Learning For Real Time Streaming Data With Kafka And Tensorflow


#DeepLearning #Tensorflow
https://www.youtube.com/watch?v=HenBuC4ATb0

✴️ @AI_Python_EN

"A machine capable of learning? That sounds wonderful." Learn how heroes of #deeplearning Yann LeCun and Ruslan Salakhutdinov first became interested in #AI:

✴️ @AI_Python_EN

For Who Have a Passion For:

1. Artificial Intelligence
2. Machine Learning
3. Deep Learning
4. Data Science
5. Computer vision
6. Image Processing

Link Group:
@DeepLearningML

We need your help to oppose the HR1044. This petition is made by the Chinese community in the US. As immigrants they believe the HR1044 will hurt the economy and skew the immigration toward one nation and only one which is India. Please Share it With your Groups and Friends.🌺

https://petitions.whitehouse.gov/petition/do-not-pass-bills-s386-or-hr1044-do-not-pass-bills-s386-or-hr1044-us-worker-and-racial-diversity

10 new posts on datahacker.rs.
Introduction to #ComputerVision using #OpenCV (#Python and #C++)

https://lnkd.in/gZtj_g6

✴️ @AI_Python_EN

Welcome to @ai_machinelearning_big_data the world of :
* #Artificial #Intelligence,
* #Deep #Learning,
* #Machine #Learning,
* #Data #Science
* #Python Programming language
* and more advanced research
links🔗 and more you wanted.
Join us and learn hot topics of Computer Science together.👇👇👇

@ai_machinelearning_big_data

Python is the main trend in programming IT and technologies .The most actual information about python programming, big data , machine learning , neural networks on channel : @pythonl

Student Guilherme Lopasso from the Springboard AI/ML course worked on a really cool project where he developed a text summarization tool by using a simple extractive approach. In particular, he used 92k articles from CNN stories as input data. Check out his blog post for more details on the project from data review, modeling to production. He also built a simple web interface which is deployed on Heroku.

It's very impressive what he has achieved so far in such short time and we're still not done yet with the course and also the project. Our next step is to use deep learning to do the text summary. #machinelearning

📝 Article: https://lnkd.in/dkF-Aj4
🔤 Code: https://lnkd.in/d8kMPPp
▶️ Demo: https://lnkd.in/dJXJT36

✴️ @AI_Python_EN

Introduction to Deep Learning

#deeplearning #datascience #artificialintelligence

https://lnkd.in/e5NpfXZ

✴️ @AI_Python_EN

Fast Estimating Pedestrian Moving State Based on Single 2D Body Pose by Shallow Neural Network.

Crossing or Not-Crossing (C/NC) problem is important to autonomous vehicles (AVs) to safely interact with pedestrians.

However, this problem setup ignores pedestrians walking along the direction of vehicles' movement (LONG).

To enhance AVs' awareness of pedestrians behavior, authors make the first step towards extending C/NC to C/NC/LONG problem and recognize them based on single body pose.

Paper: https://lnkd.in/dU8SinE
GitHub Code & JAAD dataset coming soon

#sensors #selfdrivingcars #autonomousvehicles #deeplearning

✴️ @AI_Python_EN

Facebook AI and Carnegie Mellon researchers have built Pluribus, the first AI bot to beat elite poker pros in 6 player Texas Hold’em. This breakthrough is the first major benchmark outside of 2 player games and we’re sharing specifics on how we built it.
https://ai.facebook.com/blog/pluribus-first-ai-to-beat-pros-in-6-player-poker/
✴️ @AI_Python_EN

Massively Multilingual NMT in the wild: 100+ languages, 1B+ parameters, trained using 25B+ examples. Check out our new paper for an in depth analysis:
https://arxiv.org/abs/1907.05019
#GoogleAI
✴️ @AI_Python_EN

"Large Memory Layers with Product Keys"
https://arxiv.org/abs/1907.05242

✴️ @AI_Python_EN

Great summer read: FCOS (Fully Convolutional One-Stage object detection)
https://arxiv.org/abs/1904.01355
Simpler than the already simple RetinaNet architecture, with a couple of neat tricks. Pic below from paper, probably cherry-picked 😇 but still impressive. Every orange is boxed.

✴️ @AI_Python_EN

Playing Go without Game Tree Search Using Convolutional Neural Networks.
http://arxiv.org/abs/1907.04658
✴️ @AI_Python_EN