Can artificial intelligence, in addition to recognizing the human face, understand the change of its moods?

According to studies, artificial intelligence is able to recognize people’s faces; But it does not have the same function as the human brain. The application of artificial intelligence is limited to face recognition and does not provide other useful information such as changing facial expressions.

How does artificial intelligence recognize faces?

Today, artificial intelligence is used to simultaneously recognize the face in unlocking the phone or laptop, entering social media, passing through the airport gate, etc. Artificial intelligence does this with convolutional neural networks, abbreviated as DCNN. DCNN is the main component of artificial intelligence, which was inspired by the organization of visual pathways in the brain. This organization is layered and its complexities increase by passing through each layer.

The primary layers of the DCNN structure detect simple parts of an image, such as its color and edges. The functions of this structure become more complex in each layer than the previous layer; until the last layers identify the identity of the face.

Now, the important question that arises is whether DCNN artificial intelligence, in addition to face recognition, can identify human behavior and brain mechanisms to understand more complex functions of the face, such as changing its expressions?

Recently, the Dartmouth research team collaborated with the University of Bologna and published the results of their research in the Proceedings of the National Academy of Sciences. The purpose of this research was to answer the question. In the published results, it was found that AI is not a suitable tool for understanding how to process moving faces; Because facial recognition with artificial intelligence is done only if it is static and stable. In fact, the design of artificial intelligence is such that it limits its performance to recognizing still images.

Jiahui Guo, one of the senior researchers of this study, believes:

“Scientists are trying to use deeper neural networks to understand brain functions; But the findings of recent research show that this tool has significant differences with the human brain, and at least currently it cannot function like it.”

After this research, the scientists found that although the AI ​​neural codes for faces in different DCNNs were very similar, the brain activity did not correlate well with the DCNN. These neural networks could only capture a small part of the information encoded in the brain. This shows that artificial neural networks, in their current state, are not able to understand how to process moving faces.

Conclusion

DCNNs are neural networks that form the main component of AI. Artificial intelligence can recognize the identity of people’s faces with their help. But the performance of artificial intelligence does not exceed this range; In the sense that AI is not able to identify facial expressions and changes just like the human brain. The reason for this limitation is that these neural networks are only designed to recognize static images and cannot understand the brain mechanisms that are effective in changing facial expressions. In this article, we tried to explain how to recognize faces with artificial intelligence along with its current limitations.