I'm open-minded, highly motivated in research of my interest. I'm willingly to learn new knowledge and glad to use learned knowledge to real life. During study, I tend to verify everything I learned is logical and clear. This is one of reasons makes me prefer statistics rather than deep learning approach, even the later ones have better performance on various tasks.
As for my research interest, representation learning, I believed that a good representation of inputs can help a lot on algorithm, and a carefully designed algorithm to learn such representations are critical, for a perfect representation can be applied to various tasks. One of the examples is GloVe, and Word2Vec. Besides, time series financial data is also interesting, as they are hard to predict, yet with a wide application in real world.
There are two sayings deeply touched me. The first one is from Edwards Deming, "In God we trust, others must bring data.", cited in Elements of Statistical Learning, and the other from the Gospel of John, "You shall know the truth, and the truth shall set you free".
My first appearance on academic conference
This is 2014, Hangzhou. That paper, when I look back, is not that good to be published. But it's a precious memory for me, as this conference is my first time to get in touch with professions from all over the world.
I won't tell the title of that paper :).
"Algorithm is an art. Representations are the paint. A good representation learning model transforms the data from unlovely discrete symbols, into a masterpiece of continous vectors. The preset target will be predicted, and classified, and ranked, into a portrait of accurate results. Representation Learning, where the form meets functions." - modified from an ad. in Borderlands 2.