DEEP LEARNING

Recently Google AI team presented a set of neural networks, allowing robots to be trained to move one-, two- and three-dimensional deformable objects. The code opens up opportunities for increasing the level of automation in manufacturing.

set of Google AI neural networks for teaching robots to move deformable objects
set of Google AI neural networks for teaching robots to move deformable objects
An example of a trained Transporter Network policy in action on the bag-items-1 task. Source: arxiv

About the problem

The difficulty that arises when manipulating a deformable object lies in the impossibility of completely setting its configuration. For example, to describe the location of a rigid cube in three-dimensional space, it is sufficient to indicate the position of a fixed point relative to its center, but in an object such as a cloth, the position of all its points changes relative to each…


PYTHON

📊Python & Dash📊: Guide to Better Data Visualization
📊Python & Dash📊: Guide to Better Data Visualization
Source: proglib.io

If not long ago the creation of analytical web applications required knowledge of several programming languages, today you can create a data visualization interface in pure Python. One popular tool for this has become Dash, which allows data scientists to display results in interactive web applications.

In this guide, we’ll cover:

  • How to create a Dash app
  • Main library components
  • How to customize the app style
  • How to make an application interactive
  • How to deploy an application to a remote server (using Heroku as an example)

What is Dash?

Dash is an open-source framework for building data visualization interfaces. …


DEEP LEARNING


Photo by Gilly on Unsplash

A person perceives 90% of information through sight. Clear visualization makes it easier to convey complex information. And what could be more confusing and more complicated than the stock and securities market?

Trading platforms, market forecasting agencies create tons of web-based tools for traders and analysts to interact with datasets. Technical progress is their unconditional assistant. Financial data can be visualized in various ways depending on the purpose. Let’s look at the outstanding JavaScript tools in data visualization capabilities.

1️⃣ Highcharts


DEEP LEARNING

Figure 2: Traditional and learned image codecs compared on the Kodek dataset. Source: Github

The tool is based on the PyTorch deep learning framework. The task of data compression is to minimize the size of the data and at the same time preserve information as much as possible.

More about the library

Currently, CompressAI provides:

  • Custom operations, layers, and neural network models for data compression;
  • Part of the functionality from the official TensorFlow library for compression;
  • Pretrained end-to-end compression models for image compression tasks;
  • Scripts for evaluating and comparing trained models with classical methods of compressing photo and video data

The current version of the library does not support multi-GPU training. …


DATA SCIENCE

Artwork by the author. Photo: Pexels

Python and R have long been the standard for Data Science. The essence of their opposition is that both languages ​​are great for working with statistics. While Python has clear syntax and a large number of libraries, the R language was developed specifically for the statistician, and therefore is equipped with high-quality data visualization. SQL stands out — because if the data is already in tables, then it’s more luck than a reason for frustration — and Scala — mainly due to the fact that the most popular distributed data processing framework Spark is written in it.

To conduct primary…


DEEP LEARNING

PPG Cells. Example frames per ω = 64 windows (top), and their PPG cells (bottom) consisting of raw PPG and PPG PSD, of a real video (left) and its deep fakes per generative model (rest). Source: Arxiv

Researchers at Binghampton University and Intel Corporation have developed a model that recognizes deepfakes by predicting heartbeats. The classifier uses photoplethysmogram data to recognize fake videos. An important assumption in the model is that it learns to recognize deepfakes that have been generated using a set of publicly available architectures. This imposes restrictions on the use of the model in real applications.

The approach detects fake videos with an accuracy of 97.27% and a generative model of deepfakes with an accuracy of 93.39%.

DeepFakes problem

The popularity of deepfakes has grown in recent years. Artificially generated videos of famous people are used…


DEEP LEARNING

Source: FAIR Github

FrankMocap is a new state-of-the-art neural network for 3D body and hand movement recognition that was recently developed and published by researchers at Facebook Artificial Intelligence Research (FAIR).


DEEP LEARNING

Source: Google AI Blog

Objectron is a new SOTA dataset, recently presented by Google AI, which intended to improve 3D object recognition in videos. The dataset contains 15 thousand short video clips, each containing annotation of 3D boundaries of objects. The dataset contains both real clips and synthetic ones, that is, generated based on real ones.

What is the problem

The dataset is designed to facilitate the process of training models for 3D objects on 2D image and video data.

While 2D prediction only provides 2D bounding boxes, by extending prediction to 3D, one can capture an object’s size, position, and orientation in the world, leading to a…


PRODUCTIVITY

Photo by Andreas Klassen on Unsplash

Do you do all the things time-management gurus tell you to but still end up less productive than you thought you would? There are several secrets that not everyone will tell you about. I will touch on 4 of the ones that stood out to me.

After an immense amount of research on the world’s most productive people, it becomes evident that they define productivity differently. It’s not about doing more, not about stuffing your schedule to the brim, not about getting more things done. …

Mikhail Raevskiy

Bioinformatician at Oncobox Inc. (@oncobox). Research Associate at Moscow Institute of Physics and Technology (@mipt_eng).

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store