| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 6.51 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
A privacidade e a segurança dos dados têm sido uma das principais preocupações dos utilizadores da Internet nos últimos anos. Com o aumento do uso de inteligência artificial e dos dados existentes na Internet, os utilizadores começaram a ficar preocupados com a forma como as empresas utilizavam os seus dados. Com isto, vários países começaram a desenvolver os seus próprios regulamentos de proteção de dados. Para cumprir estes regulamentos e continuar a oferecer serviços baseados na inteligência artificial, as empresas e os investigadores criaram uma técnica chamada aprendizagem federada.
A aprendizagem federada distingue-se da aprendizagem automática tradicional ao deslocar o processo de treino do servidor central para os clientes. Os clientes treinam o modelo com os seus dados locais e, em seguida, o modelo é enviado de volta para o servidor, onde os agregará num único modelo. Esta abordagem ajuda a distribuir o poder computacional e a evitar a exposição dos dados dos clientes. No entanto, há problemas ainda inexplorados pela comunidade de investigação, que dizem respeito ao lado dinâmico da aprendizagem federada.
A investigação atual parte do princípio de que os clientes estão fixos numa federação durante toda a execução, que os seus dados não mudam ao longo do tempo e que não existem outras opções para além da federação em que se encontram a treinar.
Os sistemas multiagentes, por outro lado, são um conceito que se baseia fundamentalmente em ambientes dinâmicos e que mudam com o tempo. Estes sistemas são utilizados para conceber e simular cenários realistas, incorporando características humanas, como a inteligência, a proatividade e a adaptabilidade, em entidades chamadas agentes. A sua capacidade adaptativa e comportamental que os permite seguir um objetivo pré-estabelecido faz com que se torne relevante para a aprendizagem federada, no sentido em que os clientes podem ver o seu ambiente a ser alterado, como os seus objetivos iniciais, a distribuição e a
quantidade dos seus dados, entre outros. Esta abordagem de treino colaborativo com as capacidades adaptativas e dinâmicas dos agentes pode contribuir para tornar os sistemas de aprendizagem federados mais robustos e realistas.
Esta dissertação tem como objetivo combinar ambos os conceitos através de uma ferramenta de desenvolvimento dedicada, validando a sua eficácia num cenário em que os clientes podem escolher qual o melhor modelo de federação a utilizar. A framework proposta chama-se PEAK FL e permite a criação de sistemas de aprendizagem federados baseados em agentes. Esta solução visa ajudar os utilizadores a concetualizar, a construir e a analisar as vantagens e desvantagens que este tipo de sistemas apresentam. Foram realizados vários casos de estudos para validar as suas várias componentes e os resultados mostram que, quando corretamente aplicada, esta combinação pode melhorar os resultados.
Data privacy and security have been a main concern of internet users for the last few years. With the rise of artificial intelligence applications and the amount of data available on the internet, users started to become worried about the companies using their data. Several countries started to develop their own data protection regulations to protect their users. To abide by the regulations and still offer artificial intelligence-based services, companies and researchers come up with a technique called federated learning. Federated learning distinguishes itself from traditional machine learning by moving the training process from the central server to the clients. The clients train the model with their local data and then they send the model back to the server where it will aggregate them into a single model. This process is repeated until the desired performance is reached. This approach helps to distribute the computational power and to avoid exposing the clients’ data. However, some problems were not yet fully explored by the research community, regarding the dynamic side of federated learning. Current research assumes that the clients are fixed in one federation for the entire run and that their data does not change throughout time and that there are no other options besides the federation they are in. Multi-agent systems, on the other hand, is a concept that is fundamentally based on dynamic and time-changing environments. These systems are used to design and simulate realistic scenarios by incorporating human characteristics, like, intelligence, proactiveness and goal-oriented behaviours, in these entities called agents. They can work together to achieve a common goal and adapt to the circumstances they are in. This is relevant to federated learning in the sense that the clients have a common goal, and the environment can change over time, like the data type and distribution the client has, and better federations might appear, creating the need for clients to change their behaviour throughout time, so the model performance is not reduced, and they can still achieve their desired goal. This collaborative training approach with the adaptive and dynamic agent capacities to work in unstable environments can work together to make federated learning systems more robust and realistic. This dissertation aims to combine both concepts through a dedicated development framework and its validation in a scenario where the clients can choose which federation model is better for them to use. The novel framework proposed is called PEAK-FL and allows the creation of federated learning systems based on agents. This solution aims to help the users conceptualize and build this kind of system and explore more deeply the benefits and drawbacks that these systems have. Several case studies were conducted to validate several components of the framework and the results show that, when properly applied, the integration of agents can help the clients to achieve better results.
Data privacy and security have been a main concern of internet users for the last few years. With the rise of artificial intelligence applications and the amount of data available on the internet, users started to become worried about the companies using their data. Several countries started to develop their own data protection regulations to protect their users. To abide by the regulations and still offer artificial intelligence-based services, companies and researchers come up with a technique called federated learning. Federated learning distinguishes itself from traditional machine learning by moving the training process from the central server to the clients. The clients train the model with their local data and then they send the model back to the server where it will aggregate them into a single model. This process is repeated until the desired performance is reached. This approach helps to distribute the computational power and to avoid exposing the clients’ data. However, some problems were not yet fully explored by the research community, regarding the dynamic side of federated learning. Current research assumes that the clients are fixed in one federation for the entire run and that their data does not change throughout time and that there are no other options besides the federation they are in. Multi-agent systems, on the other hand, is a concept that is fundamentally based on dynamic and time-changing environments. These systems are used to design and simulate realistic scenarios by incorporating human characteristics, like, intelligence, proactiveness and goal-oriented behaviours, in these entities called agents. They can work together to achieve a common goal and adapt to the circumstances they are in. This is relevant to federated learning in the sense that the clients have a common goal, and the environment can change over time, like the data type and distribution the client has, and better federations might appear, creating the need for clients to change their behaviour throughout time, so the model performance is not reduced, and they can still achieve their desired goal. This collaborative training approach with the adaptive and dynamic agent capacities to work in unstable environments can work together to make federated learning systems more robust and realistic. This dissertation aims to combine both concepts through a dedicated development framework and its validation in a scenario where the clients can choose which federation model is better for them to use. The novel framework proposed is called PEAK-FL and allows the creation of federated learning systems based on agents. This solution aims to help the users conceptualize and build this kind of system and explore more deeply the benefits and drawbacks that these systems have. Several case studies were conducted to validate several components of the framework and the results show that, when properly applied, the integration of agents can help the clients to achieve better results.
Description
Keywords
Federated learning Multi-agent systems Machine learning Dynamic client mobility Genetic programming