Topic: 请问有做machine learning 方面工作的吗，我想了解一下当前应用较普遍的算法。 @佛州华人论坛:佛州枫下论坛 The Rolia Forum of Florida

工作学习 / 学科技术讨论 / 请问有做machine learning 方面工作的吗，我想了解一下当前应用较普遍的算法。 -hhyang(hahay); 2011-6-1 (#6714205@0)

把问题提得细一点：decision tree learning

artificial neural networks

genetic programming

optimization under uncertainty

support vector machines

clustering

Bayesian learning

pattern recognition

constrained nonlinear programming

data mining -hhyang(hahay); 2011-6-18 {246} (#6751261@0)
When I was working on my parallel inductive logic programming in its application in data mining thesis, I should have asked the same question. -mikesmith(老猫); 2011-6-18 (#6751267@0)

I'm passionate on these problems. I've studied books on these topics, and done practical work on mathematical modelling and nonlinear programming.I am going to work on some practical problems in the field of feature extraction, probabably will need more knowledge on stochastic proramming, Baysian learning, etc. I'm also interested in the topic you mentioned, wondering if I can read your thesis? -hhyang(hahay); 2011-6-18 {252} (#6751287@0)

Thanks for ur interest. It will probably be a waste of ur time. Anyway, will PM u when I am back home -mikesmith(老猫); 2011-6-18 (#6751308@0)

thought no one would ever read it after these many years.. I am so honored... -mikesmith(老猫); 2011-6-18 (#6751469@0)

Still, not sure about which feature extraction you're talking about: (1) if feature extraction in the context of sentiment analysis, say, battery life of digital camera, you need to know NLP/IR and most importantly POS;(2) if feature selection or dimensionality reduction in the context of selecting the most important features or attributes from multiple attributes or variables, then your best bet will be PCA, SVD, NMF, etc. -renjl0810(Virtual Void); 2011-6-18 {208} (#6751512@0)

Thanks a lot for your suggestions, they are of great value to me. Could you let me know what NMF stands for, so I can look up some materials to learn?My tasks should belong to the second category, though I haven't determined attributes of my feature vector. I'm familiar with PCA,SVD, however NMF is new to me. -hhyang(hahay); 2011-6-18 {162} (#6751885@0)

Non-Negative Matrix Factorization, which is pretty new and promising algorithm. Also I just started reading a paper on MDS - Multi-Dimensional Scaling, pretty much the same thing. Should be interesting.Just no time to evaluate and compare these algorithms. -renjl0810(Virtual Void); 2011-6-18 {54} (#6751949@0)

Thanks for your information, I'll also look into both algorithms. -hhyang(hahay); 2011-6-18 (#6751988@0)
I've got some books on NMF and MDS, and I'm reading them. Looks like NMF is a hot topic. -hhyang(hahay); 2011-6-20 (#6753881@0)

如果你有 training data 和比较直接的 feature data，直接找个 black box library 扔进 SVM 就得了 -sowen(昂居居); 2011-6-20 (#6754147@0)

@Florida