Videos and lectures on MachineLearning DataScience and informatics

Ryan Urbanowicz

Perelman School of Medicine at the University of Pennsylvania

https://www.med.upenn.edu/urbslab/videos.html

Neural Path Planning: Fixed Time, Near-Optimal Path Generation via Oracle Imitation
Bency et al.: https://arxiv.org/abs/1904.11102

#Robotics #ArtificialIntelligence #MachineLearning

A Mean Field Theory of Batch Normalization
Yang et al.: https://arxiv.org/abs/1902.08129

#ArtificialIntelligence #NeuralComputing #NeuralNetworks #MachineLearning #DynamicalSystems

Spectral Inference Networks (SpIN)
Paper by Pfau et al.: https://arxiv.org/abs/1806.02215
Code: https://github.com/deepmind/spectral_inference_networks

#MachineLearning #DeepLearning #ArtificialIntelligence

A new critique of deep-learning systems that use neural nets skewers some of the current AI hype.


https://www.technologyreview.com/f/609875/the-case-against-deep-learning-hype/

SafeML ICLR 2019 Workshop
Accepted Papers: https://sites.google.com/view/safeml-iclr2019/accepted-papers

#ArtificialIntelligence #AISafety #MachineLearning
@ArtificialIntelligenceArticles

"Billion-scale semi-supervised learning for image classification"
by I. Zeki Yalniz, Hervé Jégou, Kan Chen, Manohar Paluri, Dhruv Mahajan.

Weakly-supervised pre-training + semi-supervised pre-training + distillation + transfer/fine-tuning =
81.2% twith ResNet-50,
84.8% with ResNeXt-101-32x16,
top-1 accuracy on ImageNet.

ArXiv: https://arxiv.org/abs/1905.00546

Brought to you by Facebook AI.

Original post by Yalniz Zeki: https://www.facebook.com/i.zeki.yalniz/posts/10157311492509962

@ArtificialIntelligenceArticles

#weekend_read
Paper-Title: Reinforcement Learning, Fast and Slow
#Deepmind #Cognitive_Science
Link to the paper: https://www.cell.com/action/showPdf?pii=S1364-6613%2819%2930061-0

TL;DR: [1] This paper reviews recent techniques in deep RL that narrow the gap in learning speed between humans and agents, & demonstrate an interplay between fast and slow learning w/ parallels in animal/human cognition.
[2] When episodic memory is used in reinforcement learning, an explicit record of past events is maintained for making decisions about the current situation. The action chosen is the one associated with the highest value, based on the outcomes of similar past situations.
[3] Meta-reinforcement learning quickly adapts to new tasks by learning strong inductive biases. This is done via a slower outer learning loop training on the distribution of tasks, leading to an inner loop that rapidly adapts by maintaining a history of past actions and observations.

Fast Interactive Object Annotation with Curve-GCN

Ling et al.

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


@ArtificialIntelligenceArticles

#deeplearning #machinelearning #pytorch #technology

TensorStream A library for real-time video stream decoding to CUDA memory

By Constanta: https://github.com/Fonbet/argus-tensor-stream

@ArtificialIntelligenceArticles

#ArtificialIntelligence #DeepLearning #MachineLearning

Fast AutoAugment

Lim et al.:

https://arxiv.org/abs/1905.00397
Code:
https://github.com/KakaoBrain/fast-autoaugment

#MachineLearning #ComputerVision #PatternRecognition

The Curious Case of Neural Text Degeneration

Holtzman et al.: https://arxiv.org/abs/1904.09751

#deeplearning #technology #machinelearning

Decrappification, DeOldification, and Super Resolution
By Jeremy Howard and Uri Manor: https://www.fast.ai/2019/05/03/decrappify/
#ArtificialIntelligence #DeepLearning #MachineLearning

"Assessing the Scalability of Biologically-Motivated
Deep Learning Algorithms and Architectures" https://arxiv.org/pdf/1807.04587.pdf

Here's a quick guide for everyone.


https://medium.freecodecamp.org/want-to-know-how-deep-learning-works-heres-a-quick-guide-for-everyone-1aedeca88076

Interactive Fortran Compiler

Demo: https://hub.mybinder.org/user/lfortran-web-lfortran-binder-a2ikenqm/notebooks/Demo.ipynb

#Fortran #Jupyter

Building a Silicon Brain
Computer chips based on biological neurons may help simulate larger and more-complex brain models.
https://www.the-scientist.com/features/building-a-silicon-brain-65738 https://t.iss.one/ArtificialIntelligenceArticles

The Neurons of our minds are far more functional than the modelled Neurons in Artificial Neural Networks (ANN).

Behavior from one neuron can change the behavior of another neuron; not through observation, but rather through injection of behavior #AI #DeepLearning #Neuroscience #NeuralNetworks

Source: https://buff.ly/2Fw6QH4

Using Deep Learning to Annotate the Protein Universe

Bileschi et al.: https://biorxiv.org/cgi/content/short/626507v1

#ArtificialIntelligence #Biology #DeepLearning #Technology

Microsoft launches a drag-and-drop machine learning tool

Article by Frederic Lardinois: https://techcrunch.com/2019/05/02/microsoft-launches-a-drag-and-drop-machine-learning-tool-and-hosted-jupyter-notebooks/

#ArtificialIntelligence #DeepLearning #MachineLearning

Learning to Learn How to Learn: Self-Adaptive Visual Navigation Using Meta-Learning

Wortsman et al.: https://arxiv.org/abs/1812.00971

#ArtificialIntelligence #DeepLearning #MetaLearning