paper-with-me

홈 › Papers

GradMax: Growing Neural Networks using Gradient Information

2022-01-13 · ICLR 2022 4 · Utku Evci, Bart van Merriënboer, Thomas Unterthiner, Max Vladymyrov, Fabian Pedregosa

The architecture and the parameters of neural networks are often optimized independently, which requires costly retraining of the parameters whenever the architecture is modified. In this work we instead focus on growing the architecture without requiring costly retraining. We present a method that adds new neurons during training without impacting what is already learned, while improving the training dynamics. We achieve the latter by maximizing the gradients of the new weights and find the optimal initialization efficiently by means of the singular value decomposition (SVD). We call this technique Gradient Maximizing Growth (GradMax) and demonstrate its effectiveness in variety of vision tasks and architectures.

📄 PDF Abstract BibTeX arXiv:2201.05125

Code (1)

google-research/growneuron 공식 구현 tf

Similar Papers 제목 키워드 기반

Differentiable Quality Diversity

2021-06-07 · NeurIPS 2021 12 · Matthew C. Fontaine, Stefanos Nikolaidis

Quality diversity (QD) is a growing branch of stochastic optimization research that studies the problem of generating an archive of solutions that maximize a given objective function but are also diverse with respect to …

DiversityStochastic Optimization

Bayesian Uncertainty for Gradient Aggregation in Multi-Task Learning

2024-02-06 · Idan Achituve, Idit Diamant, Arnon Netzer, Gal Chechik 외

As machine learning becomes more prominent there is a growing demand to perform several inference tasks in parallel. Running a dedicated model for each task is computationally expensive and therefore there is a great int…

Bayesian InferenceMulti-Task Learning

Momentum Gradient Descent Federated Learning with Local Differential Privacy

2022-09-28 · Mengde Han, Tianqing Zhu, Wanlei Zhou

Nowadays, the development of information technology is growing rapidly. In the big data era, the privacy of personal information has been more pronounced. The major challenge is to find a way to guarantee that sensitive …

Federated LearningPrivacy Preserving

Building Gradient Bridges: Label Leakage from Restricted Gradient Sharing in Federated Learning

2024-12-17 · Rui Zhang, Ka-Ho Chow, Ping Li

The growing concern over data privacy, the benefits of utilizing data from diverse sources for model training, and the proliferation of networked devices with enhanced computational capabilities have all contributed to t…

Federated Learning

Quantum Natural Policy Gradients: Towards Sample-Efficient Reinforcement Learning

2023-04-26 · Nico Meyer, Daniel D. Scherer, Axel Plinge, Christopher Mutschler 외

Reinforcement learning is a growing field in AI with a lot of potential. Intelligent behavior is learned automatically through trial and error in interaction with the environment. However, this learning process is often …

Multi-Armed Banditsreinforcement-learningReinforcement Learning