GithubHelp home page GithubHelp logo

raman's Introduction

Semi Supervised Learning with DCGAN on Raman Spectra Data

  1. Reproduce CNN results from RRUD database
  2. Gather large labeled database if possible
  3. Gather spectra from small set of clinically labeled urine samples
  4. Gather spectra from large unlabeled set of urine samples
  5. Build Semi Supervised DCGAN using RRUD feature bottleneck as convolutions within Discriminator, add Dense layers and Softmax mapping to classes of all clinical states plus “generated”

Review of Literature

Recent studies, such as Bispo 2016, Saatkamp 2016 and Bispo 2017, have demonstrated the feasibility of using Raman Spectroscopy (RS) to estimate the concentrations of urine metabolites such as urea, creatinine and glucose. These studies further demonstrated the feasibility of using these estimated metrics to create classification models to distinguish between healthy patients and patients with renal disease. All of these studies use similar statistical methods to analyze spectra and produce their classification models. Notably, each study uses polynomial-fit baselining to remove fluorescence followed by smoothing and normalization techniques. After baselining the spectra, the studies then apply a partial least squared (PLS) regression to generate concentration estimates before using these estimates to produce a classifier.

A recent study by Liu, Gibson, et al 2017 demonstrated the feasibility of using deep Convolutional Neural Networks (CNNs) as a universal solution for baselining and classifying Raman spectra of geological compounds. The paper demonstrates that deep CNNs are able to more accurately process and classify raw Raman spectra than the leading baselining methodologies. This opens an opportunity to apply transfer learning approaches to use the convolutional layers of the feature bottleneck as a general purpose baselining method across different Raman applications, including those in the biomedial field. Beyond the advantage of superior classification performance, this transfer learning approach has the additional benefit of potentially reducing the amount of task-specific labelled data points required to train new classifiers. If the existing feature bottleneck can fully handle the raw spectral processing, then the dense classifier layer may be able to train classifiers on the smaller labelled datasets common in biomedical applications.

raman's People

Contributors

derekkaknes avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.