An overview for using #R for validated work:

1.) Base R #Validation for #FDA: https://lnkd.in/ep8TRM8

2.) #RStudio IDE Validation: https://lnkd.in/e34FCXn

3.) Evaluating Package Stability

4.) Evaluating Package Dependencies: https://lnkd.in/eniCXgG

5.) Organizing Packages with an Internal Repository: https://lnkd.in/etSGuk4

#rstats

✴️ @AI_Python_EN

We released our new interactive annotation approach, which outperforms Polygon-RNN++ and is 10x faster.

Paper:
https://arxiv.org/pdf/1903.06874.pdf

Video:
https://www.youtube.com/watch?v=ycD2BtO-QzU …

Code:
https://github.com/fidler-lab/curve-gcn

✴️ @AI_Python_EN

A preprint for our #naacl2019 paper "Combining Sentiment Lexica with a Multi-View Variational #Autoencoder" is now online! We combine lexica with different polarity scales with a novel multi-view VAE.
https://arxiv.org/abs/1904.02839

✴️ @AI_Python_EN

PaintBot: A Reinforcement Learning Approach for Natural Media Painting

Jia et al.: https://lnkd.in/ez5Vqav

#ComputerVision #PatternRecognition #ReinforcementLearning #Painting

✴️ @AI_Python_EN

TensorFlow Vs PyTorch: Top 10 Differences Between The Two ML Libraries

#TensorFlow Vs #PyTorch #ML

✴️ @AI_Python_EN

Introduction to the math of backprop

By Deb Panigrahi: https://lnkd.in/ddtyj_U

#ArtificialIntelligence #BackPropagation #DeepLearning #NeuralNetworks

✴️ @AI_Python_EN

Music Transformer

Huang et al.: https://lnkd.in/dzHEH4E

#DeepLearning #Transformer #MachineLearning #SpeechProcessing #Music

✴️ @AI_Python_EN

Four troubling trends in Machine Learning scholarship:
1. failure to distinguish between explanation and speculation;

2. failure to identify the sources of empirical gains, e.g., emphasizing unnecessary modifications to neural architectures when gains actually stem from hyper-parameter tuning;

3. mathiness: the use of mathematics that obfuscates or impresses rather than clarifies, e.g., by confusing technical and non-technical concepts; and

4. misuse of language, e.g., by choosing terms of art with colloquial connotations or by overloading established technical terms.

https://arxiv.org/abs/1807.03341

✴️ @AI_Python_EN

Six easy ways to run your Jupyter Notebook in the cloud

By Data School: https://lnkd.in/exbAJ-S


✴️ @AI_Python_EN

Understanding Neural ODE's

Blog by Jonty Sinai: https://lnkd.in/e2SEzmZ

#artificialintelligence #machinelearning #neuralnetworks

✴️ @AI_Python_EN

Making Algorithms Trustworthy

#Algorithms

Algorithms

✴️ @AI_Python_EN

New paper & new dataset for spoken language understanding 🗣🎙🤖
Spoken language understanding (SLU) maps speech to meaning (or "intent"). (This is usually the actual end goal of speech recognition: you want to figure out what the speaker means/wants, not just what words they said.)

Paper: https://arxiv.org/abs/1904.03670
Code: https://github.com/lorenlugosch/pretrain_speech_model
Data: https://www.fluent.ai/research/fluent-speech-commands/

✴️ @AI_Python_EN

The conventional way to do SLU is to convert the #speech into text, and then convert the text into the intent. For a great example of this type of system, see this paper by alice coucke and others: https://arxiv.org/abs/1805.10190
Another approach is end-to-end SLU, where the speech is mapped to the intent through a single neural model. End-to-end SLU: -is simpler, -maximizes the actual metric we care about (intent accuracy), -and can harness info not present in the text, like prosody (e.g. sarcasm).
End-to-end #SLU is theoretically nice, but learning to understand speech totally from scratch is really hard—you need a ton of data to get it to work. Our solution: transfer learning! First, teach the model to recognize words and phonemes; then, teach it SLU.
Some people at GoogleAI and fb research have been doing some excellent work on end-to-end SLU, but without access to their datasets, it's impossible for most people to reproduce their results or do any useful research.
So we created an SLU dataset, Fluent Speech Commands, which http://Fluent.ai is releasing for free!

It's a simple SLU task where the goal is to predict the "action", "object", and "location" for spoken commands.
We hope that you find our dataset, #PyTorch code, pre-trained models, and paper useful. Even if you don't want to do SLU, the dataset can be used as a good old #classification task, adding to the list of open-source #audio datasets. Enjoy!

✴️ @AI_Python_EN

#Facebook #AI Open-Sources #PyTorch-BigGraph Tool for ‘Extremely Large’ Graphs

🌎 Open-Sources PyTorch-BigGraph Tool


✴️ @AI_Python_EN

Defeats GAN: A Simpler Model Outperforms in Knowledge Representation Learning

Heng Wang & Mingzhi Mao: https://lnkd.in/eU6THn5

#MachineLearning #ArtificialIntelligence #GenerativeAdversarialNetworks

✴️ @AI_Python_EN

"Reinforcement Learning with Attention that Works: A Self-Supervised Approach"

Manchin et al.: https://lnkd.in/exJpZDJ

#ReinforcementLearning #DeepLearning
#Visualisation

✴️ @AI_Python_EN

#Datacleaning can be especially problematic in the case of surveys.

Fieldwork companies will typically do some cleaning but, for IP reasons, are often reluctant to share the details of what they have done and on what grounds.

Cleaning is not as simple as flagging "bad" respondents. Many respondents answer some questions quite diligently but others less so. Their answers may reflect their true feelings, though, and are not necessarily "bad."

So, when conducting modeling, statisticians need to think carefully about the questions they may use for a particular model and which respondents (if any) to exclude from the modeling.

Even when the analytics will consist of simple cross tabs, it may not be safe to assume the data are "clean enough." It might be smart to have your statistician check the data before you begin your analysis.

We can also be proactive and alert our fieldwork company in advance regarding response patterns which suggest satisficing, or even include our own traps in the #questionnaire.

✴️ @AI_Python_EN