Deep learning is receiving increased attention throughout physics community as well as the real world.In this talk, after a brief introduction of deep learning, I will present two of my recent research on this technique applied to collider physics. The first part of the talk is on the possibility of strange-quark tagging, the last missing piece among quark and gluon identifications in jets. I will describe how to overcome the most difficult classification between strange and down quark jets. Neural networks feed jet images and learn features of strange jets in a supervised way. The second part is on an unsupervised learning technique called autoencoder as a tool for new physics search. The key idea of the autoencoder is that it learns to map background events back to themselves, but fails to reconstruct anomalous events that it has never encountered before. The reconstruction error can then be used as an anomaly threshold. As the first baby step, the example of finding top and gluino jets from background QCD jets will be discussed.