Titulo Estágio
Predicting computer failures from images
Áreas de especialidade
Engenharia de Software
Sistemas Inteligentes
Local do Estágio
CISUC
Enquadramento
Computers and the applications running on them fail on a frequent basis. These failures cause severe problems, such as unavailability and incorrect computations.
Predicting these failures before they occur would allow for preventive measures, potentially reducing or even preventing the impact of such failures.
Early research into this topic has used artificial intelligence models based on simple neural networks, random forests and other classical approaches. As the field of artificial intelligence advances, so should the research on how to predict failures.
Objetivo
The aim of this dissertation is to create a model capable of predicting computer failures from images constructed from the various features that represent the state of the system (e.g., CPU, memory, disk usage) and their variation along time. This model will leverage recent advances in image classification and deep learning approaches (e.g., CNNs) to achieve more robust predictions than classical approaches (e.g., random forest).
A private dataset is available to support this dissertation, which will be used to create the images and train the models.
By the end of the dissertation, it is expected that a scientific paper will be written describing the work and the results.
Plano de Trabalhos - Semestre 1
T1 (M1-M2): Perform a state-of-the-art analysis on the topics of online failure prediction, deep learning and CNNs
T2 (M3): Get acquainted with the existing dataset and prepare any script required to preprocess the data
T3 (M3-M4): Develop a script or application that produces the images that represent the system state along time
T4 (M5): Write the intermediate report
Plano de Trabalhos - Semestre 2
T5 (M6-M9): Develop, train and improve a model for failure prediction
T6 (M10): Write a scientific paper
T7 (M11): Write the final dissertation report
Condições
The dissertation will take place in the research centre CISUC (SSE group). This dissertation will be supported by a scholarship (~990€/month) during the 2nd semester.
Observações
This dissertation will be co-supervised by Prof. Henrique Madeira.
Orientador
Frederico Cerveira
fmduarte@dei.uc.pt 📩