Stanford Information Session

As an expert in the field, Dr. Zhu came to Stanford University to speak with undergraduates about how image search works, applications, and challenges. Dr. Huican Zhu is the pioneer at Google image search and is highly praise for his excellent computer vision research capability. Huican went to Google at 2000 and took responsibility of many kinds of significant Google researches such as building the AI and computer vision software for image, facial, and location detection. At Amino Capital Dr. Huican Zhu is a Founding Partner.

Dr. Zhu spoke about traditional neural networks vs. convolutional neural networks in image search. A convolutional neural network is executed using fully connected layers through local connections and tied weights followed by some form of pooling. Dr. Zhu was able to build off the students current knowledge of neural networks in order to discuss the implementation of an image search.

Deep Learning + Huge Data + Processing Power = the Future of AI

To inspire Stanford undergraduates to pursue their interest in this field, Dr. Zhu provided some applications of image search. Some of the applications that students found most interesting included: self driving cars and allowing blind people to see through a camera and image search software to alert the user by vibrations.

One main challenge of image search and computer vision is training the model to be as accurate and specific as possible. We discussed the complications of image searching for autonomous car software in difficult weather conditions.

This theme of big data and image search relates back to Amino Capital and Dr. Zhu’s influence at the firm. They invested in Orbeus, ultimately acquired by Amazon, which uses deep learning for automated facial, object and scene recognition,image-to-text analysis and video indexing. Amino Capital looks for investments in big data and using technology to solve large scale problems.

This article was contributed by Rachel Ahn, a Campus Ambassador at Stanford University.

To learn more about Amino Capital, please contact hello@aminocapital.com