In past, Google has launched remarkable AI-based projects whether it’s AlphaGo that beat’s the world’s board game champions or the Google Duplex which is a completely automated system that places calls like humans.
Now the researchers at Google Brain have developed an artificial intelligence (AI) system that has created its own child. What’s more, the original AI has trained its creation to such a high level that it outperforms every other human-built AI system.
This system is called AutoML, an artificial intelligence (AI) that’s capable of generating its own AIs. The engineers explain that there aim with AutoML is to make the process of designing machine learning models easier and automate.
“Our approach can design models that achieve accuracies on par with state-of-art models designed by machine learning experts (including some of our own team!),” the post notes. They even add that the AI children have certain design features that seem to be of no clear use to their own researchers.
The Googles automated the design of machine learning models by using reinforcement learning. The machine learns to do so through evolutionary algorithms and reinforcement learning algorithms.
The AutoML acts as a controller neural network which we can perhaps call the parent that develops a child AI network for a specific task. A particular task is given to the child and the feedback given to the parent.
This feedback is then used to tell the controller how to improve its proposals for the next round. The process is then repeated thousands of times, generating new architectures, testing them, and giving that feedback to the controller to learn from.
The child created by AI is named as NASNet, Google designed it, as a system that can do object detection which included people, cars, traffic lights, handbags, backpacks, among others, in a video in real-time.
NASNet later gets tested on the ImageNet image classification and COCO object detection data sets, which the Google researchers call “two of the most respected large-scale academic data sets in computer vision”.
NASNet outperformed all other computer vision systems in this test. Eventually, NASNet was 82.7 percent accurate at predicting images correctly.
This is 1.2 percent better than any previously published results, and the system is also 4 percent more efficient, with a 43.1 percent mean Average Precision (mAP).
Additionally, a less computationally demanding version of NASNet outperformed the best similarly sized models for mobile platforms by 3.1 percent.
The Google researchers acknowledge that NASNet could prove useful for a lot of applications and have open-sourced the AI for inference on image classification and object detection.
“We hope that the larger machine learning community will be able to build on these models to discuss multitudes of computer vision problems we have not yet imagined,” said the researchers, who have open-sourced NASNet so it could be used for computer vision applications.”
While there are many possible uses of AutoML and NASNet, there are also ethical issues related to AI.
For instance, what if AutoML creates AI systems at such a speed that the society simply cannot keep up with them or what if the AI parent passes down unwanted biases to its child.
To keep all these things in human control, it is very important to carry out more strict regulations and enhanced ethical standards to prevent the use of AI for malicious purposes.
More in AI :