With the highly hyped entry of ChatGPT, much attention has been paid to artificial intelligence (AI), how it can be trained, and its boundaries. Numerous learning methods are part of artificial intelligence, such as Machine Learning and Deep Learning. More precisely, machine learning and deep learning are both subfields of AI, and both involve training computer systems to learn from data. However, there are some important differences separating both notions.
A focus on Machine Learning:
Machine learning is part of artificial intelligence and is based on an algorithm that relies on human feedback and requires structured data. The algorithm adapts according to the new information captured. It is a method of teaching computers to learn data without being explicitly programmed. It involves training models to feed the system with structured and categorized data, and those models will help the system to make predictions or decisions about new data. Machine learning algorithms can be divided into three main categories: supervised learning, unsupervised learning, and reinforcement learning (1).
What about Deep Learning?
Deep Learning, on the other hand, doesn’t require structured data. It is a subset of machine learning, that works with the use of a multi-layer neural networks (hence the term “deep”). Deep learning use much more data than machine learning, and deepl learning models are capable of learning from large and complex data sets. Deep learning is mostly used in IT security, customer support, content creation, speech assistant, etc. (2).
Therefore, deep learning is a specialized form of machine learning that involves training large neural networks with many layers, but all tasks can be carried out by both systems. Deep learning and machine learning are not used in the same areas of applications, as deep learning requires more IT resources (more data, cost more, etc.).
Both deep learning and machine learning involve legal issues including privacy, intellectual property, accountability and liability. An important aspect concerns the use of these technologies under the GDPR.
Machine Learning & Deep Learning under the GDPR
The General Data Protection Regulation (GDPR) is a European Union regulation that governs the collection, processing, and storage of personal data. Machine learning and deep learning are both concerned by the GDPR rules, as they are part of artificial intelligence.
Under GDPR, users and individuals have the right to know how their data is being used and have the right to control how their data is collected, processed, and stored.
Therefore, companies and organizations that use machine learning must be transparent about how they are using personal data, and most importantly, obtain the explicit consent from users before collecting and using their data.
In addition, GDPR requires companies and organizations to take all necessary security measures to anticipate any unauthorized access, disclosure, or loss of any personal data.
Furthermore, under the principle of “Privacy by design“, the design of Machine Learning that exploits personal data must take into account security measures in order to guarantee the privacy of users, and to minimize any risk.
Overall, GDPR puts in place restrictions and obligations regarding the use and exploitation of personal data in machine learning, which in particular helps to guarantee the privacy and rights of users. Companies that design these tools must comply with the GDPR, or risk facing various sanctions from the CNIL.
- (1) https://datascientest.com/quelle-difference-entre-le-machine-learning-et-deep-learning
- (2) https://www.ionos.com/digitalguide/online-marketing/search-engine-marketing/deep-learning-vs-machine-learning/