The contribution inspired by the brain: Researchers bring Ki human vision first appeared at the online magazine Basic Thinking. You can start the day well every morning via our newsletter update.
Researchers have developed a new method to teach AI models-like vision. The so-called LP Convolution is based on the brain to improve the accuracy of image labeling systems.
Researchers from the Institute for Basic Science (IBS), the Yonsei University and the Max Planck Institute have a new one method Developed that enables AI systems like human-like vision. The so-called LP Convolution is intended to improve the accuracy and efficiency of image recognition systems and at the same time reduce the computing effort.
The idea behind it: machines should learn to see as a brain just as flexible, focused and robustly. Conventional BILD-KIs have so-called Convolutional Neural Networks (CNNS). These analyze pictures with rigid filters – usually three times three pixels large windows – and recognize patterns.
This works well in many cases, but not always. Especially when important image information is distributed or complex. Alternatives such as Vision Transformers are powerful, but also extremely calculating and are often not suitable for everyday life.
LP Convolution: AI learns to see human-like
Instead of rigid filters, the LP Convolution is based on adaptable views. They behave like a rubber band that can stretch or contract as required. The special thing: The flexible filters are modeled on the connections in the human brain. They mathematically orient themselves on the Gausschen distribution. This ensures that we not only perceive the center of the field of vision sharply, but also remains sensibly integrated.
The result is a AI that can not only recognize relevant image content faster, but is also more robust compared to disorders. In practice, the method could enable more precise decisions in areas such as medical technology, autonomous driving or robotics.
Better, more flexible, human
In tests, the LP Convolution exceeded many existing models-with a significantly lower arithmetic effort. The neuronal activity inside the AI is similar to that of a real mouse in the brain. This proximity to biological perception makes the method particularly effective.
This applies to both research and future applications in everyday life. Because AI models could not only rigid and analyze visual information with the approach using patterns, but actually “understand”.
Also interesting:
- What is actual machine learning?
- The current Chatgpt models in comparison
- Kluger-Hans effect: When artificial intelligence becomes a danger
- Artificial intelligence: What are neural networks?
The contribution inspired by the brain: Researchers bring Ki human vision first appeared on basic thinking. Follow us too Google News and Flipboard Or subscribe to our update newsletter.
As a Tech Industry expert, I find the concept of teaching AI human vision incredibly fascinating and promising. The human brain is incredibly complex and powerful, so it makes sense to look to it for inspiration when developing artificial intelligence systems. By better understanding how our brains process visual information, researchers can potentially improve AI’s ability to understand and interpret images in a more human-like way.
This research has the potential to revolutionize a wide range of industries, from healthcare to autonomous vehicles to security systems. By teaching AI to see and interpret the world more like humans do, we can unlock new possibilities for how AI can assist and enhance our daily lives.
However, it’s important to proceed with caution and ethical considerations in mind. As AI continues to advance and become more human-like in its capabilities, we must ensure that it is being used responsibly and ethically. Transparency, accountability, and fairness must be prioritized in the development and deployment of AI systems inspired by the human brain.
Overall, I am excited to see how this research progresses and the potential impact it could have on the future of AI technology. By combining the power of human intelligence with the capabilities of artificial intelligence, we can create truly groundbreaking innovations that benefit society as a whole.
Credits