Federated Machine Learning : The future of AI in a privacy-obsessed world
Greetings !!!. I am Paritosh, working as a Data Scientist in India. Feel free to connect with me , this is my linkedin profile . Feel Free to ask any queries.
Data is key factor to improve performance of our machine learning model. But we know that the data, our model encounters some times have tagged as sensitive data. So it is not a best practice to upload all the users data into server and train it. It leads to the destruction of user’s privacy. But thanks to federated learning , a secure way of training our model within the user device. In this user data is collected and set training environment of the model within the user’s device , and the trained results are sent to the server and then encrypted ,making it more secure without leaving the traces of the data it is trained. Generally group of devices are selected to perform training. You may get a doubt of disturbing performance of the device but these tasks are generally performed when the device is in charging or some time at rest.
What is Federated Learning:
a. It is a Framework.
b. An n number of data owners who want to build a common model will feed their data respectively. All data is put together (i.e. , the sum of data is taken), hence individual data of each other owner is not exposed.
Trending AI Articles:
Major Problems faced in AI Today:
a. Traditional data processing models in AI often involves simple data transaction models, therefore we will be unclear of future uses of the models and we may violate the laws of GDPR(General Data Protection Regulation). Hence data is in the form of isolate islands.
b. Strengthening of Data Privacy and Security.
Solution to the Problems : Federated Machine Learning
a. Proposed by Google.
b. The main idea is to build machine learning models based on datasets that are distributed across multiple devices while preventing data leakage.
c. Standard machine learning approaches require centralizing the training data on one machine or datacentre , whereas federated learning helps to learn the shared prediction model while keeping all training data on the device.
How is Privacy Protected in Federated Machine Learning:
This requires security models and analysis to provide meaningful privacy guarantees. The different privacy techniques are used:
a. Secure Multiparty Computation ( SMC ) : SMC naturally involve multiple parties and provide security proof in a well-defined simulation framework to guarantee complete zero knowledge , that is , each party knows nothing expect its input and output.
b. Differential Privacy / K- anonymity : The methods of differential privacy , K- anonymity , and diversification involve in adding noise to the data or using generalization methods to obscure certain sensitive attributes until the third party cannot distinguish the individual , thereby making the data impossible to be restored to protect user privacy.
c. Homomorphic Encryption : It is adopted to protect user data privacy through exchange under the encryption mechanism during machine learning.
Therefore , there is little possibility of leakage at the raw data level.
Categorization of Federated Learning :
a. Horizontal Federated learning : Horizontal federated Learning , or sample-based federated learning , is introduced in the scenarios that datasets share the same feature space but different in samples.
b. Vertical Federated Learning : Vertical Federated learning or Feature-based federated learning is applicable to the cases that two datasets share the sample ID space but differ in feature space.
c. Federated Transfer Learning (FTL) : FTL applies to the scenarios that the two datasets differ not only in samples but also in feature space.
APPLICATIONS :
As an innovative modelling mechanism that could train a united model on data from multiple parties without compromising privacy and security of those data , federated learning has a promising application in sales , financial and many other industries , in which data can not be directly aggregated for training ML models due to factors such as intellectual property rights, privacy protection , and Data security.
CONCLUSION :
In recent years , the isolation of data and the emphasis on data privacy are becoming the next challenges for artificial intelligence , but federated learning has brought us new hope. It would break the barriers between industries and established a community where data and knowledge could be shared together with safety , and the benefits would be fairly distributed according to the contribution of each participant. The bows of Artificial Intelligence would be finally be brought to every corner of our lives.
This is my Second blog post. I will be happy to receive any constructive feedback which will help me improve and motivate me to write more blogs. If you have any questions or help needed feel free to reach out to me by
Email : paritoshkr30@gmail.com
Linkedin: https://www.linkedin.com/in/paritosh-kumar-b8ab61141/
Clap ,if you like this post. Thank you !!!!
Don’t forget to give us your 👏 !
