​How We Finished 3rd and Still Won a Data Science Competition

🔗 How We Finished 3rd and Still Won a Data Science Competition
We had a lot of work to do — even after the final scores were published.

​Visual Recognition using Graphs - Towards Data Science

🔗 Visual Recognition using Graphs - Towards Data Science
An overall package for understanding and implementing graph convolutions for semantic segmentation.

​Text Classification by XGBoost & Others: A Case Study Using BBC News Articles

🔗 Text Classification by XGBoost & Others: A Case Study Using BBC News Articles
Comparative study of different vector space models & text classification techniques like XGBoost & others

​«Умное» видеонаблюдение: какой будет жизнь под камерами с искусственным интеллектом

Камеры наблюдают за нами почти беспрерывно, но толку в этом мало. Если человек не анализирует картинку, камера остаётся просто прибором, производящим терабайты часов малопригодного стрима. Альтернатива — снабдить камеру ИИ-инструментами. И вот такая система видеонаблюдения способна будет заменить спящего перед монитором охранника, в офисе — босса и в супермаркете — маркетолога. Рассказываем, как именно.
https://habr.com/ru/company/toshibarus/blog/458094/

🔗 «Умное» видеонаблюдение: какой будет жизнь под камерами с искусственным интеллектом
Камеры наблюдают за нами почти беспрерывно, но толку в этом мало. Если человек не анализирует картинку, камера остаётся просто прибором, производящим терабайты...

🎥 Accelerating High-Resolution Weather Models with Deep-Learning Hardware
👁 1 раз ⏳ 1769 сек.
In this video from PASC 2019, Sam Hatfield from the University of Oxford presents: Accelerating High-Resolution Weather Models with Deep-Learning Hardware.

"The next generation of weather and climate models will have an unprecedented level of resolution and model complexity, and running these models efficiently will require taking advantage of future supercomputers and heterogeneous hardware. In this paper, we investigate the use of mixed-precision hardware that supports floating-point operations at double

🎥 Секция Data: Что делать, если стандартной функциональности MILib недостаточно
👁 1 раз ⏳ 2159 сек.
Фреймворк Spark широко распространен в наши дни в крупных компаниях и имеет много приложений. Библиотека Spark MLlib позволяет решать классические задачи машинного обучения. Но что делать, если стандартного функционала не хватает? В докладе мы рассмотрим подходы, которые позволяют добавить функционал к существующему фреймворку и в то же время сохранить production ready свойство Spark'а.

​IBM/pytorchpipe

🔗 IBM/pytorchpipe
PyTorchPipe (PTP) is a component-oriented framework for rapid prototyping and training of computational pipelines combining vision and language - IBM/pytorchpipe

​10 Python image manipulation tools.

An overview of some of the commonly used Python libraries that provide an easy and intuitive way to transform images.

https://towardsdatascience.com/image-manipulation-tools-for-python-6eb0908ed61f

🔗 10 Python image manipulation tools - Towards Data Science
An overview of some of the commonly used Python libraries that provide an easy and intuitive way to transform images.

​Everything you need to know about TensorFlow 2.0
Keras-APIs, SavedModels, TensorBoard, Keras-Tuner and more.

https://hackernoon.com/everything-you-need-to-know-about-tensorflow-2-0-b0856960c074

🔗 Everything you need to know about TensorFlow 2.0 - Hacker Noon
Keras-APIs, SavedModels, TensorBoard, Keras-Tuner and more.

​TensorFlow is dead, long live TensorFlow!

https://hackernoon.com/tensorflow-is-dead-long-live-tensorflow-49d3e975cf04

🔗 TensorFlow is dead, long live TensorFlow! - Hacker Noon
If you’re an AI enthusiast and you didn’t see the big news this month, you might have just snoozed through an off-the-charts earthquake…

🎥 Deep Learning @twitter: Twitter meets Tensorflow - Cibele Halasz
👁 1 раз ⏳ 1694 сек.
Twitter has a long story in Machine Learning. Over the past year, we have been transitioning to a new chapter of this story by migrating our ML platform from Lua Torch to Tensorflow. This talk will be focused on the Machine Learning framework we have been developing on top of Tensorflow. Ultimately, we want to give you an idea of how we are doing Machine Learning at Twitter’s scale.
--
Cibele is a Software Engineer at Twitter Cortex, where she builds Twitter’s deep learning platform. Prior to working at Tw

🎥 Friday Seminar Series: Hui Jiang - Understanding Deep Learning Theory
👁 1 раз ⏳ 5240 сек.

🎥 CS480/680 Lecture 15: Deep neural networks
👁 1 раз ⏳ 5466 сек.

🎥 Machine Learning LIVE #1 - Learn Python, Numpy, OpenCV, Pandas & Grab Deep Learning T-Shirt
👁 1 раз ⏳ 3622 сек.
Update - The Submission Link for Pokemon Challenge with Fresh leaderboard will be released tomorrow!
-----------------------------------------------------------------------------------------------------------------------------------------------

Coding Blocks presents you an exclusive webinar on Machine Learning with extensive coverage of Python and Libraries like Numpy, Pandas, Matplotlib, OpenCV. The second part of the webinar will be premiered tomorrow in which you will build the Pokemon Classifier!

A

​Deep Coupled-Representation Learning for Sparse Linear Inverse Problems with Side Information

Authors: Evaggelia Tsiligianni, Nikos Deligiannis

Abstract: …the computational complexity. In this paper, we consider the recovery of a target signal with the aid of a correlated signal, the so-called side information (SI), and propose a deep unfolding model that incorporates SI.
https://arxiv.org/abs/1907.02511

🔗 Deep Coupled-Representation Learning for Sparse Linear Inverse Problems with Side Information
In linear inverse problems, the goal is to recover a target signal from undersampled, incomplete or noisy linear measurements. Typically, the recovery relies on complex numerical optimization methods; recent approaches perform an unfolding of a numerical algorithm into a neural network form, resulting in a substantial reduction of the computational complexity. In this paper, we consider the recovery of a target signal with the aid of a correlated signal, the so-called side information (SI), and propose a deep unfolding model that incorporates SI. The proposed model is used to learn coupled representations of correlated signals from different modalities, enabling the recovery of multimodal data at a low computational cost. As such, our work introduces the first deep unfolding method with SI, which actually comes from a different modality. We apply our model to reconstruct near-infrared images from undersampled measurements given RGB images as SI. Experimental results demonstrate the superior performance of the

​How We Finished 3rd and Still Won a Data Science Competition

🔗 How We Finished 3rd and Still Won a Data Science Competition
We had a lot of work to do — even after the final scores were published.

This AI Makes The Mona Lisa Come To Life

#DeepFake

🎥 This AI Makes The Mona Lisa Come To Life
👁 1 раз ⏳ 266 сек.
❤️ Check out Weights & Biases here and sign up for a free demo:
https://www.wandb.com/

📝 The paper "Few-Shot Adversarial Learning of Realistic Neural Talking Head Models" is available here:
https://arxiv.org/abs/1905.08233v1
https://www.youtube.com/watch?v=p1b5aiTrGzY

🙏 We would like to thank our generous Patreon supporters who make Two Minute Papers possible:
313V, Alex Haro, Andrew Melnychuk, Angelos Evripiotis, Anthony Vdovitchenko, Brian Gilman, Bruno Brito, Bryan Learn, Christian Ahlin, Christoph Jad

​Deep Learning Recommendation Model for Personalization and Recommendation Systems:

An implementation of a deep learning recommendation model (DLRM) The model input consists of dense and sparse features. The former is a vector of floating point values. The latter is a list of sparse indices into embedding tables, which consist of vectors of floating point values. The selected vectors are passed to mlp networks denoted by triangles, in some cases the vectors are interacted through operators (Ops).https://github.com/facebookresearch/dlrm

🔗 facebookresearch/dlrm
An implementation of a deep learning recommendation model (DLRM) - facebookresearch/dlrm

​Секция JavaScript: Машинное обучение в JavaScript Библиотеки и решения

🔗 Секция JavaScript: Машинное обучение в JavaScript Библиотеки и решения
Известный факт, что Machine learning - это сфера разработки, которая помогает решать всё большее и большее количество прикладных задач. Но o ML с использованием JavaScript известно пока мало. В докладе я постараюсь раскрыть эту тему. Речь пойдёт об инструментах и возможностях, которыми мы можем пользоваться уже сейчас.

🎥 Свёрточные нейронные сети: dropout, image / data augmentation. Борьба с переобучением.
👁 1 раз ⏳ 881 сек.
⚡⚡⚡ На прошлом уроке мы разработали небольшую свёрточную нейронную сеть для классификации изображений элементов одежды в оттенках серого цвета из набора данных FASHION MNIST. Мы убедились на практике в том, что наша небольшая нейронная сеть может классифицировать поступающие на вход изображения с достаточно высокой точностью. Однако в реальном мире нам предстоит работать с изображениями высокого разрешения и различных размеров. Одним из замечательных преимуществ СНС является тот факт, что они могут так же х