内容简介:Data science has gained a tremendous popularity in recent years. More and more businesses see the potential of data to create value. Thus, there has been an increase in demand for data scientists which encourage lots of people to start a career in the fiel
Including data sources and example projects.
Jul 11 ·6min read
Data science has gained a tremendous popularity in recent years. More and more businesses see the potential of data to create value. Thus, there has been an increase in demand for data scientists which encourage lots of people to start a career in the field of data science. There is absolutely no lack of resources to learn data science nowadays. I did not count but I’m pretty sure there are more than a thousand online certificates related to data science. When books, podcasts, and youtube videos are added to the pile, it becomes a huge resource collection to consume.
It is good to have a variety of resources to learn. However, after you learn the basics and current tools and software packages, it is time to challenge yourself with projects. What independent projects add to your skillset cannot be achieved with online certificates. I have a detailed post on why you should start doing projects. Here it is if you want to take a look at it.
In this post, I will list 5 project ideas along with how you can find relevant datasets. I will also give a link to an example project which can help you get started.
1. Image Classification
One of the areas that make use of deep learning is computer vision. There are numereous applications of neural networks in this field such as image recognition, detecting or generating fake images and videos. Neural networks are also widely used in the health care industry. For instance, neural networks have proved to be successful in cancer detection using x-rays. An image classification project would be your first step into this broad field.
You are likely to use convolutional neural networks (CNNs) for image recognition. CNNs are commonly used in data science domain especially for computer vision and image classification tasks. Images consist of pixels which are represented with numbers. In the convolution layer of CNNs, filters (or feature detectors) are applied to the image to extract distinctive features of the image by preserving the spatial relationships among pixels.
Caltech101 dataset contains lots of images of many different objects. It is a great dataset to train and test a CNN. Many thanks to the community who prepared and let us use this dataset.
Here is a walk-through of creating a basic image classification model:
以上就是本文的全部内容,希望本文的内容对大家的学习或者工作能带来一定的帮助,也希望大家多多支持 码农网
猜你喜欢:本站部分资源来源于网络,本站转载出于传递更多信息之目的,版权归原作者或者来源机构所有,如转载稿涉及版权问题,请联系我们。