Blockchain as a new technology of public trust?14 January 2018
TensorFlow – a free machine learning library for everyone.15 January 2018
Google answered this question. Their artificial intelligence created a more perfect AI design for face recognition than people did so far. It is called NASNet. Developed by people from Google Brain, the neural network project called AutoML aims to create an artificial intelligence that itself will create further artificial intelligence.
AutoML creates better than a human
AutoML has created a new vision algorithm that far exceeds the most modern models. During tests, it recognized faces better than all models created by human. To test NASNet, Google used it to classify the ImageNet image and the COCO object detection data set, which it describes as “the two most-respected academic data sets in a computer vision.”
According to NASNet scientists, it had 82.7 per cent accuracy in forecasting images in the ImageNet validation set. This is a 1.2% better result than all previously published results, and the system is 4% more efficient. In addition, the computational less-demanding NASNet exceeded the best models with similar sizes for mobile platforms by 3.1 per cent.
ML and AI – threat or help for humanity?
Machine learning gives many artificial intelligence systems the ability to perform specific tasks. Although the concept behind this is quite simple – the algorithm learns, transferring a lot of data – the process requires a huge amount of time and effort.
When it comes to NASNet, accurate and efficient computer vision algorithms are highly sought after due to the number of potential applications. According to scientists, they can be used to create advanced robots with artificial intelligence support or to help visually impaired people recover their eyesight, as suggested by one of the researchers. They could also help designers improve the technology of self-propelled vehicles. The faster an autonomous vehicle recognizes objects in its path, the faster it can react to them, thus increasing the safety of such vehicles. Google researchers admit that NASNet can be useful for a wide range of applications and has an open AI probe to request image classification and object detection.
The system is “open source”, therefore Google team hopes that “a larger community of machine learners will be able to build on these models to solve many computer problems that we have not yet imagined.”
However, the creation of artificial intelligence that can build artificial intelligence raises some concerns. For example, what prevents a parent from transmitting unwanted prejudices to his child? What if AutoML creates systems so fast that society cannot keep up?
It is not difficult to understand how NASNet can be used in automated surveillance systems in the near future, perhaps earlier than regulations could be introduced to control such systems. One of the people who are opposed to the uncontrolled development of AI is the general manager of Tesla, Elon Musk. Together with a group of people, he called for the regulation of artificial intelligence, because it is “a fundamental risk to the existence of human civilization”.