SMOTE as a strategy for balancing dataset

This article solves the following challenge: 

Imbalanced dataset classification performs poorly

Use docker-compose to easily start multiple docker containers

In order to easily start up multiple docker instances with one command, one could use docker-compose. This command line tool lets user specify a docker compose file, which uses YAML to define which containers need to be started and how they will be connected.
The exemplary definition for one docker container to be started would look like this (note that the indentaion is lost, because whitespace is trimmed)

- ...

See changes that were made by user in latex document

The core problem is to compare two latex generated documents and highlight differences between them. The important part here is that a simple *.tex document may have dependencies to other documents or included .tex files (i.e. each chapter/section may be stored as a separate file). Therefore, one needs to find a way to first combine and build both projects (i.e. produce output), then compare them and then produce a single pdf file that highlights differences.

Imbalanced dataset classification performs poorly

Imbalanced datasets can have tremendeous negative impact on training a classification model. Is there a way in Azure ML to deal with these kind of challenge?

Starting multiple docker containers and let them communicate

When developing microservice-like architectures it is reasonable to assume that docker is used for deploying the individual services. In order to test if the services worke together properly, especially on a small scale, on can start a docker container for each service. Since each of the services gets its own network stack created by docker, they can by default not communicate with each other.
It would be nice to have the possiblity to simply start all services at once and give them the possibility to communicate with each other using the hostnames of the services.

Tile database is broken

There might be the case, that after windows update did automatically install new updates which are deployed when shutting windows down, the database where the tiles in the start menue are configured is broken.

To restore the tile database of the start menue:

  • First close the Windows Explorer in the task manager
  • Pressing [Ctrl] + [Shift] + [Del] opens the task manager.
    Press More details to expand the task manager view.
    Find Windows-Explorer and close the task by [Right click] + [Close task].

  • Open powershell

Windows 10 Start menue disappeared

In Windows 10, after starting the PC, the start menue was gone, nothing happened when clicking the windows sign in the bottom left corner. No reboot helped.

Integrating Dehaze module with Presets

The popular Dehaze Function is released to the Creative Cloud Version, Lightroom CC. Since Lightroom 6 is the standalone version, new features in minor releases are shipped only with the next standalone major release Ligthroom 7.

When receiving minor release updates for Lightroom 6, the sources are installed anyway when updating to the latest minor release but the user interface does not offer the dehaze module.

Dehaze module not visible in Lightroom 6

Having the latest pc version of Lightroom 6 installed and updated it to 6.6, the announced "Dehaze module" cannot be found. Since this is a popular way to improve photos with a lot of haze in it some way to integrate this function would be great.

Use some tools that support agile development

I'd recommend to use some tools for agile development to create a self-organized and self-directed work team. Agile methods advocates adaptive planning and flexible response to changes which could help with your problem.


Subscribe to TechScreen Revised RSS