惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

人人都是产品经理
人人都是产品经理
Google DeepMind News
Google DeepMind News
博客园 - 【当耐特】
量子位
博客园 - 司徒正美
爱范儿
爱范儿
Hugging Face - Blog
Hugging Face - Blog
博客园 - 聂微东
Jina AI
Jina AI
J
Java Code Geeks
腾讯CDC
大猫的无限游戏
大猫的无限游戏
V
Visual Studio Blog
I
InfoQ
D
Docker
Recent Announcements
Recent Announcements
MongoDB | Blog
MongoDB | Blog
博客园 - Franky
宝玉的分享
宝玉的分享
G
Google Developers Blog
GbyAI
GbyAI
Y
Y Combinator Blog
有赞技术团队
有赞技术团队
H
Help Net Security

Keep the gradient flowing

Policy Gradients Part 1: The REINFORCE Estimator On the Link Between Optimization and Polynomials, Part 6. Optimization Nuggets: Stochastic Polyak Step-size, Part 2 Optimization Nuggets: Stochastic Polyak Step-size On the Convergence of the Unadjusted Langevin Algorithm The Russian Roulette: An Unbiased Estimator of the Limit Notes on the Frank-Wolfe Algorithm, Part III: backtracking line-search On the Link Between Optimization and Polynomials, Part 5 Optimization Nuggets: Implicit Bias of Gradient-based Methods Optimization Nuggets: Exponential Convergence of SGD On the Link Between Optimization and Polynomials, Part 4 On the Link Between Optimization and Polynomials, Part 3 On the Link Between Optimization and Polynomials, Part 2 On the Link Between Polynomials and Optimization, Part 1 How to Evaluate the Logistic Loss and not NaN trying Notes on the Frank-Wolfe Algorithm, Part II: A Primal-dual Analysis Three Operator Splitting Notes on the Frank-Wolfe Algorithm, Part I Optimization inequalities cheatsheet A fully asynchronous variant of the SAGA algorithm Hyperparameter optimization with approximate gradient Lightning v0.1 scikit-learn-contrib, an umbrella for scikit-learn related projects. SAGA algorithm in the lightning library On the consistency of ordinal regression methods Holdout cross-validation generator IPython/Jupyter notebook gallery PyData Paris - April 2015 Data-driven hemodynamic response function estimation Plot memory usage as a function of time
Initial implementation of the query system
Fabian Pedregosa · 2009-06-12 · via Keep the gradient flowing

I sent some patches to sympy-patches with an initial implementation of the query system. You can check it out by pulling from my branch: git pull http://fa.bianp.net/git/sympy.git master into your sympy repo. Some examples of what you can do (sample isympy session): In [1]: query(x, positive=True) Returns None, as we do not know whether x is positive or not. `` In [2]: query(abs(x), positive=True) Out[2]: True`` because abs() is always positive. Because exp() is always positive, the following should also be True: `` In [3]: query(exp(x), positive=True)`` but why then does it return None ?. Well, it simply is not True that exp() is always positive, it is always positive for real values, but SymPy does not assume that x is real, so you would have to specify that. This is now done with the keyword assumptions: `` In [5]: query(exp(x), positive=True, assumptions=Assume(x, real=True)) Out[5]: True`` As you can see, now assumptions are independent objects and are not tied to symbols any more. For more examples, see the file sympy/query/tests/test_query.py still in the TODO list: - support for global assumptions - solve more complex implications, like query(x, positive=True, assumptions=Assume(x, even=True)) where it should build the chain of implications even => integer => rational => real. This chain of implications currently stops at rational for efficiency reasons, because the number of facts grows in each step which makes the number of possible paths grow exponentially.