Bjoux2
commented
7 years ago a recomeded paper is 'Deep learning in bioinformatics' by Seonwoo Min, Byunghan Lee and Sungroh Yoon.
there are two kinds of representation of gene data. one is expert-designed features for specail tasks, such as 1-mer, 2-mer, 3-mer composition, second structure. the other is features which extract from raw gene data. the raw gene data form is not frequently used.
As the paper says "DNNs (deep neural networks) are renowned for their suitability in analyzing ighdimensional data. Given that bioinformatics data are typically complex and high-dimensional, DNNs have great promise for bioinformatics research. We believe that DNNs, as hierarchical representation learning methods, can discover previously unknown highly abstract patterns and correlations to provide insight to better understand the nature of the data. However, it has occurred to us that the capabilities of DNNs have not yet fully been exploited. Although the key characteristic of DNNs is that hierarchical features are learned solely from data, human-designed features have often been given as inputs instead of
raw data forms. We expect that the future progress of DNNs in bioinformatics will come from nvestigations into proper ways to encode raw data and learn suitable features from them."
Hay.Nowdays, I am interesting in Biomedical Signal Processing by deep learning. I am confused by the biosignal sequence to extract data for learning model.So that I had to study the papers your shared with some hope. THX for your shared papers. Could you share your experience in processing the sequence data like gene data?