The Experts below are selected from a list of 26133 Experts worldwide ranked by ideXlab platform
Salman A Avestimehr - One of the best experts on this subject based on the ideXlab platform.
-
fednas federated deep learning via neural architecture search
2020Co-Authors: Murali Annavaram, Salman A AvestimehrAbstract:Federated Learning (FL) has been proved to be an effective learning framework when data cannot be Centralized due to privacy, communication costs, and regulatory restrictions. When training deep learning models under an FL setting, people employ the predefined model architecture discovered in the Centralized Environment. However, this predefined architecture may not be the optimal choice because it may not fit data with non-identical and independent distribution (non-IID). Thus, we advocate automating federated learning (AutoFL) to improve model accuracy and reduce the manual design effort. We specifically study AutoFL via Neural Architecture Search (NAS), which can automate the design process. We propose a Federated NAS (FedNAS) algorithm to help scattered workers collaboratively searching for a better architecture with higher accuracy. We also build a system based on FedNAS. Our experiments on non-IID dataset show that the architecture searched by FedNAS can outperform the manually predefined architecture.
-
towards non i i d and invisible data with fednas federated deep learning via neural architecture search
arXiv: Learning, 2020Co-Authors: Murali Annavaram, Salman A AvestimehrAbstract:Federated Learning (FL) has been proved to be an effective learning framework when data cannot be Centralized due to privacy, communication costs, and regulatory restrictions. When training deep learning models under an FL setting, people employ the predefined model architecture discovered in the Centralized Environment. However, this predefined architecture may not be the optimal choice because it may not fit data with non-identical and independent distribution (non-IID). Thus, we advocate automating federated learning (AutoFL) to improve model accuracy and reduce the manual design effort. We specifically study AutoFL via Neural Architecture Search (NAS), which can automate the design process. We propose a Federated NAS (FedNAS) algorithm to help scattered workers collaboratively searching for a better architecture with higher accuracy. We also build a system based on FedNAS. Our experiments on non-IID dataset show that the architecture searched by FedNAS can outperform the manually predefined architecture.
Murali Annavaram - One of the best experts on this subject based on the ideXlab platform.
-
fednas federated deep learning via neural architecture search
2020Co-Authors: Murali Annavaram, Salman A AvestimehrAbstract:Federated Learning (FL) has been proved to be an effective learning framework when data cannot be Centralized due to privacy, communication costs, and regulatory restrictions. When training deep learning models under an FL setting, people employ the predefined model architecture discovered in the Centralized Environment. However, this predefined architecture may not be the optimal choice because it may not fit data with non-identical and independent distribution (non-IID). Thus, we advocate automating federated learning (AutoFL) to improve model accuracy and reduce the manual design effort. We specifically study AutoFL via Neural Architecture Search (NAS), which can automate the design process. We propose a Federated NAS (FedNAS) algorithm to help scattered workers collaboratively searching for a better architecture with higher accuracy. We also build a system based on FedNAS. Our experiments on non-IID dataset show that the architecture searched by FedNAS can outperform the manually predefined architecture.
-
towards non i i d and invisible data with fednas federated deep learning via neural architecture search
arXiv: Learning, 2020Co-Authors: Murali Annavaram, Salman A AvestimehrAbstract:Federated Learning (FL) has been proved to be an effective learning framework when data cannot be Centralized due to privacy, communication costs, and regulatory restrictions. When training deep learning models under an FL setting, people employ the predefined model architecture discovered in the Centralized Environment. However, this predefined architecture may not be the optimal choice because it may not fit data with non-identical and independent distribution (non-IID). Thus, we advocate automating federated learning (AutoFL) to improve model accuracy and reduce the manual design effort. We specifically study AutoFL via Neural Architecture Search (NAS), which can automate the design process. We propose a Federated NAS (FedNAS) algorithm to help scattered workers collaboratively searching for a better architecture with higher accuracy. We also build a system based on FedNAS. Our experiments on non-IID dataset show that the architecture searched by FedNAS can outperform the manually predefined architecture.
Avestimehr Salman - One of the best experts on this subject based on the ideXlab platform.
-
Towards Non-I.I.D. and Invisible Data with FedNAS: Federated Deep Learning via Neural Architecture Search
2021Co-Authors: He Chaoyang, Annavaram Murali, Avestimehr SalmanAbstract:Federated Learning (FL) has been proved to be an effective learning framework when data cannot be Centralized due to privacy, communication costs, and regulatory restrictions. When training deep learning models under an FL setting, people employ the predefined model architecture discovered in the Centralized Environment. However, this predefined architecture may not be the optimal choice because it may not fit data with non-identical and independent distribution (non-IID). Thus, we advocate automating federated learning (AutoFL) to improve model accuracy and reduce the manual design effort. We specifically study AutoFL via Neural Architecture Search (NAS), which can automate the design process. We propose a Federated NAS (FedNAS) algorithm to help scattered workers collaboratively searching for a better architecture with higher accuracy. We also build a system based on FedNAS. Our experiments on non-IID dataset show that the architecture searched by FedNAS can outperform the manually predefined architecture.Comment: accepted to CVPR 2020 workshop on neural architecture search and beyond for representation learning. Code is released at https://fedml.a
He Chaoyang - One of the best experts on this subject based on the ideXlab platform.
-
Towards Non-I.I.D. and Invisible Data with FedNAS: Federated Deep Learning via Neural Architecture Search
2021Co-Authors: He Chaoyang, Annavaram Murali, Avestimehr SalmanAbstract:Federated Learning (FL) has been proved to be an effective learning framework when data cannot be Centralized due to privacy, communication costs, and regulatory restrictions. When training deep learning models under an FL setting, people employ the predefined model architecture discovered in the Centralized Environment. However, this predefined architecture may not be the optimal choice because it may not fit data with non-identical and independent distribution (non-IID). Thus, we advocate automating federated learning (AutoFL) to improve model accuracy and reduce the manual design effort. We specifically study AutoFL via Neural Architecture Search (NAS), which can automate the design process. We propose a Federated NAS (FedNAS) algorithm to help scattered workers collaboratively searching for a better architecture with higher accuracy. We also build a system based on FedNAS. Our experiments on non-IID dataset show that the architecture searched by FedNAS can outperform the manually predefined architecture.Comment: accepted to CVPR 2020 workshop on neural architecture search and beyond for representation learning. Code is released at https://fedml.a
Guzay Pasaoglu Kilanc - One of the best experts on this subject based on the ideXlab platform.
-
a decision support tool for the analysis of pricing investment and regulatory processes in a deCentralized electricity market
Energy Policy, 2008Co-Authors: Guzay Pasaoglu Kilanc, Ilhan OrAbstract:After the liberalization of the electricity generation industry, capacity expansion decisions are made by multiple self-oriented power companies. Unlike the Centralized Environment, decision-making of market participants is now guided by price signal feedbacks and by an imperfect foresight of the future market conditions (and competitor actions) that they will face. In such an Environment, decision makers need to better understand long-term dynamics of the supply and demand sides of the power market. In this study, a system dynamics model is developed, to better understand and analyze the deCentralized and competitive electricity market dynamics in the long run. The developed simulation model oversees a 20-year planning horizon; it includes a demand module, a capacity expansion module, a power generation module, an accounting and finance module, various competitors, a regulatory body and a bidding mechanism. Many features, singularities and tools of deCentralized markets, such as; capacity withholding, enforced divestment, long-term contracts, price-elastic demands, incentives/disincentives, are also incorporated into the model. Public regulators and power companies are potential users of the model, for learning and decision support in policy design and strategic planning. Results of scenario analysis are presented to illustrate potential use of the model.