This week, me and my roommate participate in the kaggle competition. This one is proposed by otto group and it is to predict the distribution of class_i given an item with 93 features. The visualization can be found here.

We start from using stacked auto-encoder and multilayer perceptron but the cross validation only reach around 80% correctness.

Today,we adapt Gradient Boosting Machine( one of the ensemble method), I hope this will squeeze a few percent of the correctness. It’s still running. I’ll keep posted.

If someday I wake up but I don’t have that courage to fight, I know I need to change something, something that wakes up my inner voice, woke up my startup dream. To fight for the benefit for everyone through cutting edge technology, to fight to help everyone save even a little bit of time because I improve the efficiency of the system, to fight to fundamentally improve the quality of people’s life because I come up with an great idea and successfully implement it.

Stay hungry, Stay foolish.There are a lot of things to do and improve. My time is limited given I will be cleared away not so far from now. I want to ride on this wave in deep learning “revolution” to achieve the state of the art results which will be utilized to change people’s life.

Computer scalability and efficiency gives me a lot of thinking in my real life about how to be good when start a company. All this efficiency and scalability that I learned in computer science came across my mind during a shower.

How we could scale things up? In Hadoop, we can use these mid-level even low-level hardware to form cluster to cooperate on some task obeying a master node, in high performance computing/ parallel computing, we carefully design these architecture and algorithm to try to keep every thread/cores at the same page during computing. In GPGPU computing, the parallel computing even happens on a lot of weak threads but still outperform even high-end quadro-core CPU.

All these make me think, it’s not about startup. It’s not about you can perform a really great job in one field, it’s about you first have a vision with strong field knowledge as a backend and you can hire people even “weak” people to perform better task under you “code”/instructions within an architecture. All that matters in the end is your vision, your culture(architecure), your instruction in specific task(code).