AI Benchmark App to be Released for Android

The application tests the motherboard, processor and RAM, and then issues a number indicating the effectiveness of the AI ​​on the device
27 July 2018   743

Specialists in the field of computer vision from the Swiss Higher Technical School of Zurich have created an application that assesses the performance of smartphones for working with artificial intelligence.

The results of the test will be useful for researchers of AI, manufacturers of components and developers of Android. Thanks to the information received, they will be able to learn and correct the shortcomings of the devices in order to improve the efficiency of their work with AI.

AI Benchmark can be downloaded on Google Play and run on any smartphone with Android 4.1 and higher. The application tests the motherboard, processor and RAM, and then issues a number indicating the effectiveness of the AI ​​on the device.

When testing AI Benchmark evaluates the ability of the smartphone to edit high-resolution images, recognize objects in photos and classify them. In addition, the system tests the algorithms used in autopilot for cars.

Summarizing the results presented on the project's website, the researchers came to the following conclusions:

  • Qualcomm - theoretically can give good results, but not enough drivers;
  • Huawei - quite outstanding results;
  • Samsung - there is no acceleration support, but powerful processors;
  • Mediatek - good results for middle-class devices.

The final rating of smartphones and their hardware platforms is presented below:

Rating of Android in AI Benchmark
Rating of Android in AI Benchmark

One of the creators of AI Benchmark Andrey Ignatov said that the application was developed for about three months. The idea of ​​its creation arose because of the lack of information about the limitations of the use of modern AI on smartphones. This is due to the fact that currently all algorithms work remotely on servers, not on the device, except for some pre-installed applications.

Experts are sure that in the future AI technologies will be no less important than a good camera in a smartphone, so they want to actively participate in the development of this field.

MIT CSAIL to Fight AI Bias

As reported, bias in AI leads to poor search results or user experience
19 November 2018   55

A team of scientists from the Laboratory of Informatics and Artificial Intelligence MIT has published a paper dedicated to the fight against the misconceptions that arise in neural networks in the learning process. The main attention is paid to the problem of preserving the accuracy of predicted AI results.

Since scientists have been trying to cope with the problem of discriminatory misconceptions of AI for the first year, there are traditional methods of operations in this area. Usually, for correcting training, a certain amount of information is added to the data set, which allows the neural network to obtain more accurate data on a particular sample.

Thus, in one experiment, the AI ​​should have noted the expected level of income of individuals in the presented selection. As a result of a discriminatory misconception that appeared in the process of learning, AI twice as often marked men as individuals with high incomes. Increasing the number of female profiles in a training dataset allowed us to reduce the error by 40%.

The problem with traditional methods is that the data sets prepared in this way do not reflect the actual distribution of the population. This increases the fallacy of predictions issued by AI.

In their work “Why Is My Classifier Discriminatory?” Scientists offer several possible solutions to the problem. They believe that increasing the size of the training dataset without changing the proportions of the represented gender, social and racial groups will allow the AI ​​to cope with discriminatory errors independently. According to the researchers, the collection of additional information from the same source that provided the initial data packet will avoid covariant bias.

This method can be costly, as it will have to pay for the work of specialists marking up additional data. However, researchers are confident that in many cases such costs will be justified.

The second option is clustering groups of the population most vulnerable to discrimination and the subsequent separate processing of these clusters with the introduction of additional variables. Scientists suggest using this method when obtaining additional data is difficult or impossible.