Application of Machine Learning in Artificial Intelligence
Within the last few decades, the application of machine learning, and artificial intelligence has become a very high point to talk about in society. Furthermore, machine learning which is a rapidly growing technology day by day has become the most discussed topic in the world of technology today. We use machine learning in our everyday lives without even realizing it, such as Google Maps, Google Assistant, Alexa, and many other real-world machine learning applications including automated language translation, medical diagnosis, stock market trading, image recognition, and so on (Javatpoint, 2018).
When all the above applications of machine learning are considered, image recognition can be identified as the most popular and widely used application of machine learning today. Objects, digital images, people, locations, and other things are all identified using image recognition. As a common use case factor of image recognition, automatic friend tagging suggestion which Facebook uses comes the first place when talking about image recognition. Facebook, for example, has a feature that suggests automatic friend tags (Keshari, 2020) where Facebook identifies and recognizes the person no matter how they look. When a person uploads a picture of himself, Facebook automatically detects the person and suggests tagging the person with their given name. The major technology behind this mesmerize function is the face detection and face recognition algorithms of machine learning. This feature is focused on the "Deep Face" Facebook initiative, that is in charge of facial recognition and individual identification in photos (Salama Abdelminaamid et al., 2020). So how does Facebook use machine learn in Artificial Intelligence?
Among the government agencies around the world, image and facial recognizing systems are in the rage to seek to automate services to keep track of their citizens. Consider the case where a photograph of an individual is visible anywhere, and the person can be recognized in photographs or videos from public sensor cameras. However, Facebook has recognized this as a source of concern and has invented a method to circumvent this technology for scientific purposes.
Face de-identification technology (Gafni, Wolf and Taigman, 2019), designed by three Facebook AI researchers, alter the facial expression in streaming video to the point that facial recognition algorithms are unable to match what they see in the videos with images of us stored on their servers. FAIR (Facebook AI Research) has created a cutting-edge “de-identification” method that operates on camera, including a live stream. It operates through using machine learning to change the main physical expressions of a visual object in real-time, trying to mislead an image recognition device into falsely recognizing the object (Statt, 2019). Similar features, such as the form of a person's mouth movements or the shape of their pupils, are altered. This can be used for both pre-recorded and live recording, according to Facebook.
This technique can make it much more reasonable to use video recordings of individuals to train AI systems, which usually need many samples to recognize how to replicate the feedback they are given. These Artificial intelligence systems can be conditioned by invading the privacy of test subjects by rendering people's faces unrecognizable.
Facebook has no intentions to incorporate it into any of its services, according to the publication. Face detection is used by Facebook to recognize users in posted images for easy identification and to notify users when they appear in other people's photos. Last month, the corporation decided to switch image recognition off by default, which is a minor step toward protecting its users' confidentiality.
Despite these challenges, image recognition is one of the many interesting applications of machine learning that can be integrated with artificial intelligence to apply modern technologies only if these techniques are used properly for the benefit of humans.
References
Gafni, O., Wolf, L. and Taigman, Y., 2019. Live Face De-Identification in Video.
Ilyam, M., 2020. Artificial Intelligence in Facebook. Discover 7 ways they use. [online] Available at: <https://itchronicles.com/.../7-revealing-secrets-how.../> [Accessed 6 Mar. 2021].
Javatpoint, 2018. Applications of Machine Learning - Javatpoint. [online] Available at: <https://www.javatpoint.com/applications-of-machine-learning> [Accessed 6 Mar. 2021].
Keshari, K., 2020. Top 10 Applications of Machine Learning | Daily Life Applications | Edureka. [online] Available at: <https://www.edureka.co/blog/machine-learning-applications/> [Accessed 6 Mar. 2021].
Salama Abdelminaamid, D., Almansori, A.M., Taha, M. and Badr, E., 2020. A deep facial recognition system using computational intelligent algorithms. [online] Available at: <https://doi.org/10.1371/journal.pone.0242269>.
Statt, N., 2019. Facebook researchers trained AI to fool facial recognition systems - The Verge. [online] Available at: <https://www.theverge.com/.../facebook-ai-facial...> [Accessed 6 Mar. 2021].
0 Comments