Papers with Code 아카이브
Second-order methods
태스크의 하위 분류와 논문·코드·평가·기법·데이터셋을 탐색하세요.
- 하위 태스크
- 0
- 평가표
- 0
- 보존된 평가 기록
- 0
Use second-order statistics to process data.
코드
관련 논문과 평가 결과에 연결된 구현입니다. 출처에서 확인된 저자 구현에는 공식 코드 표시가 있습니다.
https://github.com/IST-DASLab/M-FAC
공식 코드M-FAC: Efficient Matrix-Free Approximations of Second-Order Information
https://github.com/IssamLaradji/ssn
공식 코드Fast and Furious Convergence: Stochastic Second Order Methods under Interpolation
https://github.com/LOG-postech/Sassha
공식 코드SASSHA: Sharpness-aware Adaptive Second-order Optimization with Stable Hessian Approximation
https://github.com/RemiLELUC/ConditionedSGD
공식 코드Asymptotic Analysis of Conditioned Stochastic Gradient Descent
https://github.com/aakashlahoti/sliqn
공식 코드https://github.com/amirgholami/adahessian
공식 코드ADAHESSIAN: An Adaptive Second Order Optimizer for Machine Learning
https://github.com/amirgholami/hessianflow
공식 코드Large batch size training of neural networks with adversarial training and second-order information
https://github.com/blupblupblup/doppler-signatures-generation
공식 코드Near out-of-distribution detection for low-resolution radar micro-Doppler signatures
https://github.com/blupblupblup/near-ood-doppler-signatures
공식 코드Near out-of-distribution detection for low-resolution radar micro-Doppler signatures
https://github.com/brianbullins/lissa_code
공식 코드Second-Order Stochastic Optimization for Machine Learning in Linear Time
https://github.com/chaoyanghe/MiLeNAS
공식 코드MiLeNAS: Efficient Neural Architecture Search via Mixed-Level Reformulation
https://github.com/f-dangel/singd
공식 코드Structured Inverse-Free Natural Gradient: Memory-Efficient & Numerically-Stable KFAC
https://github.com/f-dangel/sirfshampoo
공식 코드Can We Remove the Square-Root in Adaptive Gradient Methods? A Second-Order Perspective
https://github.com/felix-petersen/isaac
공식 코드ISAAC Newton: Input-based Approximate Curvature for Newton's Method
https://github.com/fkunstner/limitations-empirical-fisher
공식 코드Limitations of the Empirical Fisher Approximation for Natural Gradient Descent
https://github.com/flgoyens/nonlinear-matrix-recovery
공식 코드Nonlinear matrix recovery using optimization on the Grassmann manifold
https://github.com/ghliu/snopt
공식 코드https://github.com/gmatilde/SGN
공식 코드On the Promise of the Stochastic Generalized Gauss-Newton Method for Training DNNs
https://github.com/google-research/google-research
공식 코드https://github.com/google-research/google-research/tree/master/dp_transfer
공식 코드https://github.com/hseung88/nysact
공식 코드NysAct: A Scalable Preconditioned Gradient Descent using Nystrom Approximation
https://github.com/hsivan/fosi
공식 코드https://github.com/ist-daslab/efcp
공식 코드https://github.com/jarekduda/sgd-ogr-hessian-estimator
공식 코드https://github.com/jmdvinodjmd/LIBS2ML
공식 코드Stochastic Trust Region Inexact Newton Method for Large-scale Machine Learning
https://github.com/jvictormata/adasub
공식 코드AdaSub: Stochastic Optimization Using Second-Order Information in Low-Dimensional Subspaces
https://github.com/jxbz/agd
공식 코드Automatic Gradient Descent: Deep Learning without Hyperparameters
https://github.com/lixilinx/psgd_tf
공식 코드Online Second Order Methods for Non-Convex Stochastic Optimizations
https://github.com/mohammad-mozaffari/mkor
공식 코드MKOR: Momentum-Enabled Kronecker-Factor-Based Optimizer Using Rank-1 Updates
https://github.com/mohmdelsayed/hesscale
공식 코드Revisiting Scalable Hessian Diagonal Approximations for Applications in Reinforcement Learning
https://github.com/mtkresearch/exact_GN_revNN
공식 코드https://github.com/myingysun/sgdph
공식 코드SGD with Partial Hessian for Deep Neural Networks Optimization
https://github.com/neelmishra/gauss-newton-based-minimax-solver
공식 코드A Gauss-Newton Approach for Min-Max Optimization in Generative Adversarial Networks
https://github.com/noobyzy/FedZO
공식 코드Communication-Efficient Stochastic Zeroth-Order Optimization for Federated Learning
https://github.com/p-wol/GroupedNewton
공식 코드Adapting Newton's Method to Neural Networks through a Summary of Higher-Order Derivatives
출처: Papers with Code · CC BY-SA 4.0. 아카이브 자료와 새로 수집한 논문은 별도로 관리합니다.