There are many active research projects accessing and applying shared ADNI data. Use the search above to find specific research focuses on the active ADNI investigations. This information is requested annually as a requirement for data access.
Principal Investigator | |
Principal Investigator's Name: | JONATAN REYES |
Institution: | CONCORDIA UNIVERSITY |
Department: | SCHOOL OF ENGINEERING AND COMPUTER SCIENCE |
Country: | |
Proposed Analysis: | We plan to use the ADNI dataset in our research to evaluate the predictive performance of deep learning models operating under a federated learning environment, especially when the training data is non-IID. We have formulated an aggregation algorithm that makes use of the second raw moment of the stochastic gradient to compute the weighted average of the parameters of distributed models. Preliminary results have shown a performance increase with MNIST, Fashion-MNIST, and CIFAR-10, which motivate us to examine further the feasibility of our averaging strategy with imbalanced multicentric data. For example, in ADNI, the imbalance between mild cognitive impairment (MCI) cases and Alzheimer's disease (AD) cases makes it suitable for our study. |
Additional Investigators | |
Investigator's Name: | MARTA KERSTEN-OERTEL |
Proposed Analysis: | We plan to use the ADNI dataset in our research to evaluate the predictive performance of deep learning models operating under a federated learning environment, especially when the training data is non-IID. We have formulated an aggregation algorithm that makes use of the second raw moment of the stochastic gradient to compute the weighted average of the parameters of distributed models. Preliminary results have shown a performance increase with MNIST, Fashion-MNIST, and CIFAR-10, which motivate us to examine further the feasibility of our averaging strategy with imbalanced multicentric data. For example, in ADNI, the imbalance between mild cognitive impairment (MCI) cases and Alzheimer's disease (AD) cases makes it suitable for our study. |