Shanghai Jiao Tong University's AI Can Recognize Faces
References: arxiv.org & thenextweb
Researchers from Shanghai Jiao Tong University have developed a facial recognition AI system that is able to predict men who are criminals with astounding accuracy. Though the similar, pseudoscientific field of phrenology that emerged in the 19th century has been discredited, the Shanghai Jiao Tong University AI shows that such research may not have been entirely off base.
To train the AI, the researchers fed the program over 1,500 images of male faces and told it which of these were criminals and which were not. They then fed the system close to 200 new images, but this time withheld the statuses of the men pictured. Using a neural network, the computer was able to predict whether the unknown man was a criminal with 89.5 percent accuracy -- not accurate enough to engender a 'Minority Report' like future, but far beyond being random chance.
To train the AI, the researchers fed the program over 1,500 images of male faces and told it which of these were criminals and which were not. They then fed the system close to 200 new images, but this time withheld the statuses of the men pictured. Using a neural network, the computer was able to predict whether the unknown man was a criminal with 89.5 percent accuracy -- not accurate enough to engender a 'Minority Report' like future, but far beyond being random chance.
Trend Themes
1. Facial Recognition AI - Creating more accurate and sophisticated facial recognition AI systems to predict behaviors and characteristics.
2. Criminal Prediction Technology - Developing innovative technologies that can predict criminal tendencies or behavior based on facial features.
3. Neural Network Analysis - Advancing the capabilities of neural networks to accurately analyze and make predictions based on complex data.
Industry Implications
1. Artificial Intelligence - Leveraging advancements in AI to develop powerful predictive systems that can be applied in various fields such as law enforcement, security, and risk assessment.
2. Law Enforcement - Exploring the potential of criminal-predicting AI technologies to aid law enforcement agencies in identifying potential threats and preventing crime.
3. Data Analysis - Harnessing the power of neural networks and advanced data analysis techniques to uncover patterns and insights from large datasets for various applications, including behavioral analysis.
2.8
Score
Popularity
Activity
Freshness
