Davi Frossard

Personal Page

VGG in TensorFlow

Files Model weights - vgg16_weights.npz TensorFlow model - vgg16.py Class names - imagenet_classes.py Example input - laska.png To test run it, download all files to the same folder and run python vgg16.py Introduction VGG is a convolutional neural network model proposed by K. Simonyan and A. Zisserman from the University of Oxford in the paper “Very Deep Convolutional Networks for Large-Scale Image Recognition” . The model achieves 92.

Classification

Introduction The problem of classification consists in assigning an observation to the category it belongs. That means, for instance, taking a picture of a handwritten digit and correctly classifying which digit (0-9) it is, matching pictures of faces to whom they belong or classifying the sentiment in a text. In general, when dealing with classification we use supervised learning (when we have an annotated training set from which we can learn our model - as we did up until this point).

TensorFlow

Introduction TensorFlowâ„¢ is an open source machine learning library for Python (and C++) initially developed by the Google Brain Team for research and released under the Apache 2.0 Open Source License on November 9th, 2015 . It is currently used by Google in their speech recognition, Gmail, Google Photos, Search services and recently adopted by the DeepMind team . Features Though TensorFlow was built with deep learning in mind, its framework is general enough so that we can also implement clustering methods, graphical models, optimization problems and others.

Multiple Linear Regression

This post is a continuation of Linear Regression. Introduction In multiple linear regression we extend the notion developed in linear regression to use multiple descriptive values in order to estimate the dependent variable, which effectively allows us to write more complex functions such as higher order polynomials ($y = \sum_{i_0}^{k} w_ix^i$), sinusoids ($y = w_1 sin(x) + w_2 cos(x)$) or a mix of functions ($y = w_1 sin(x_1) + w_2 cos(x_2) + x_1x_2$).

Linear Regression with NumPy

Introduction Linear regression is a method used to find a relationship between a dependent variable and a set of independent variables. In its simplest form it consist of fitting a function $ \boldsymbol{y} = w.\boldsymbol{x}+b $ to observed data, where $\boldsymbol{y}$ is the dependent variable, $\boldsymbol{x}$ the independent, $w$ the weight matrix and $b$ the bias. Illustratively, performing linear regression is the same as fitting a scatter plot to a line.

Under construction

The site is currently under construction. Come back later for more content.