I think Dimensionality reduction through clustering is well

You may use clustering, principle component analysis and a weighted graph based reduction techniques are also used now a day. It also depends upon the characteristics of the datasets which you are using. You may follow the below papers for your help,

https://academic.oup.com/bib/article/17/4/628/2240645/Dimension-reduction-techniques-for-the-integrative

https://www.knime.org/files/knime_seventechniquesdatadimreduction.pdf

https://people.cs.pitt.edu/~milos/courses/cs2750-Spring04/lectures/class20.pdf

http://www.sciencedirect.com/science/article/pii/S0957417417300593

Principal Component Analysis has proven efficient.

Thank you dears 

 i know about the PCA but i asked about the newest techniques that can be used now a day

Metric Multidimensional Scaling (MDS) is well.