“机器学习与人工智能”的版本间差异
跳到导航
跳到搜索
无编辑摘要 |
(→降维) |
||
(未显示同一用户的21个中间版本) | |||
第1行: | 第1行: | ||
* artificial neural networks |
* artificial neural networks (ANN) |
||
: 气体的冷却 deepCool, deepHeat, and deepMetal: arXiv:1901.01264 |
: 气体的冷却 deepCool, deepHeat, and deepMetal: arXiv:1901.01264 |
||
: 对quench的机制进行排序 [https://arxiv.org/pdf/1601.01258.pdf] |
|||
* unsupervised Random Forest |
* unsupervised Random Forest |
||
:找到SDSS中最怪异的星系 [https://arxiv.org/abs/1611.07526] |
:找到SDSS中最怪异的星系 [https://arxiv.org/abs/1611.07526] |
||
: https://arxiv.org/abs/2007.08530 |
|||
* 恒星形成历史 [https://arxiv.org/abs/1903.10457] |
* 恒星形成历史 [https://arxiv.org/abs/1903.10457] |
||
第14行: | 第16行: | ||
:arXiv1901.02877 1908.04318 |
:arXiv1901.02877 1908.04318 |
||
==图像相关== |
|||
*merger |
*merger |
||
:REALSIM [https://ui.adsabs.harvard.edu/abs/2019MNRAS.490.5390B/abstract] |
:REALSIM [https://ui.adsabs.harvard.edu/abs/2019MNRAS.490.5390B/abstract] |
||
*Galaxy morphology network [https://arxiv.org/abs/2006.14639] |
*Galaxy morphology network [https://arxiv.org/abs/2006.14639] |
||
* Morpheus |
|||
:https://ui.adsabs.harvard.edu/abs/2020ApJS..248...20H/abstract |
|||
*贾鹏的方法 |
|||
:https://arxiv.org/abs/2002.09211 |
|||
*deblend |
|||
:Residual Dense Neural network (RDN) [https://arxiv.org/abs/2109.09550] |
|||
:sextractor做deblend [https://github.com/esheldon/sxdes] |
|||
:weaklense中的deblend [https://github.com/LSSTDESC/WeakLensingDeblending] |
|||
*图像生成 |
|||
: [http://arxiv.org/abs/2203.11956] StyleGAN 生成的图片真实度最高,生成的图片可以增加星系样本的多样性,使得网络更具有泛化能力。 |
|||
==算法== |
|||
* 贝叶斯方法中的Simulation-based inference (SBI; also known as “likelihood-free” inference):基于神经元的正向建模 [http://arxiv.org/abs/2203.07391] |
|||
* OPTICS [https://www.biaodianfu.com/optics.html] DBSCAN的一种扩展 |
|||
* SHAPLEY:游戏模型中的一种判断某个因素重要性的算法,arXiv 2309.06784 |
|||
===降维=== |
|||
*降维:tSNE |
|||
*降维:MDS Multidimensional Scaling, [https://zhuanlan.zhihu.com/p/141434196] |
|||
*数据内禀维度:https://arxiv.org/pdf/2404.02962.pdf |
|||
===回归=== |
|||
*THE SPARSE REGRESSION METHODOLOGY,arXiv:2101.02986 |
|||
:用一组基函数去拟合某个参数,同时惩罚基函数的系数及项数,从而抑制过拟合 |
|||
==软件包== |
|||
*[[rapids]] |
|||
*[[scikit-learn]] |
|||
*astroML [https://www.astroml.org/astroML-notebooks/] |
|||
*[[jax]] |
|||
*[[pytorch]] |
|||
*[[tensorflow]] |
|||
* [[pyro]] |
|||
==教程== |
|||
*https://github.com/WillKoehrsen/Machine-Learning-Projects |
2024年4月6日 (六) 14:48的最新版本
- artificial neural networks (ANN)
- 气体的冷却 deepCool, deepHeat, and deepMetal: arXiv:1901.01264
- 对quench的机制进行排序 [1]
- unsupervised Random Forest
- 找到SDSS中最怪异的星系 [2]
- https://arxiv.org/abs/2007.08530
- 恒星形成历史 [3]
- 基于数值模拟的结果,监督学习
- Convolutional Neural Networks(CNN)
- Extremely Randomised Trees
- 高斯过程回归 (Gaussian Process Regressor)
- Scikit-Learn machine learning package for Python
- arXiv1901.02877 1908.04318
图像相关
- merger
- REALSIM [4]
- Galaxy morphology network [5]
- Morpheus
- 贾鹏的方法
- deblend
- 图像生成
- [9] StyleGAN 生成的图片真实度最高,生成的图片可以增加星系样本的多样性,使得网络更具有泛化能力。
算法
- 贝叶斯方法中的Simulation-based inference (SBI; also known as “likelihood-free” inference):基于神经元的正向建模 [10]
- OPTICS [11] DBSCAN的一种扩展
- SHAPLEY:游戏模型中的一种判断某个因素重要性的算法,arXiv 2309.06784
降维
- 降维:tSNE
- 降维:MDS Multidimensional Scaling, [12]
- 数据内禀维度:https://arxiv.org/pdf/2404.02962.pdf
回归
- THE SPARSE REGRESSION METHODOLOGY,arXiv:2101.02986
- 用一组基函数去拟合某个参数,同时惩罚基函数的系数及项数,从而抑制过拟合
软件包
- rapids
- scikit-learn
- astroML [13]
- jax
- pytorch
- tensorflow
- pyro