ML Script

Welcome to my blog.

Beside my role as architect for Cloud and Big Data at American Express i am also excited about Cryptocurrencies, Quantitative Trading and especially Machine Learning.

my linkedin profile

My Current ML Stack

install Tensorflow on GPU
install Jupyter Notebook

Libraries

Algorithms

Activation Functions

  • Sigmoid or Logistic   f(x) = 1 / 1 + exp(-x)
  • Tanh — Hyperbolic tangent  f(x) = 1 — exp(-2x) / 1 + exp(-2x)
  • ReLu -Rectified linear units if x < 0  { R(x) = 0}       if x >= 0  { R(x) = x}

Initialisation

  • Gaussian
  • GlorotUniform
  • Xavier
  • LeCunUniform

Hosted Services

Mobile

Conferences

Training

Data Sets

Books

Papers

Machine Learning Process

  • Problem Definition
  • Analyze Data
  • Prepare Data
  • Feature Scaling
  • Evaluate Algorithms
  • Train Model
  • Tune Model
  • Serve Model

Example: Forex Workflow

  • Load tick data in CSV format
  • Import into MT4
  • Build time series of Candle Bars or Heikin-Ashi Candle Bars
  • Calculate Indicators
  • Export to MongoDB
  • Import from MongoDB  into Pandas Frame
  •  Standardize to mean of 0 and standard deviation of 1
  • Lag candles and indicators to create prediction window
  • Split datasets into training, testing, validation
  • Train model in Keras
  • Export model
  • Convert model to IOS metal
  • Serve model on IOS

Leave a Reply

Your email address will not be published. Required fields are marked *