Deep Learning Algorithms
What does Deep Learning mean?
Most machine learning algorithms work well on datasets that have up to a few hundred features, or columns. However, an unstructured dataset like one from an image has such a large number of features that this process becomes cumbersome or completely unfeasible. A single 800-by-1000-pixel image in RGB color has 2,400,000 features – far too many for traditional machine learning algorithms to handle, which try and take in all the information at once.
Deep learning algorithms learn progressively more about the image as it goes through each neural network layer. Early layers learn how to detect low-level features like edges, and subsequent layers combine features from earlier layers into a more holistic representation. For example, a middle layer might identify edges to detect parts of an object in the photo like a leg or a branch, while a deep layer will be able to detect the full object like a dog or a tree.
Why is Deep Learning important?
The ability to process large numbers of features makes deep learning very powerful when dealing with unstructured data. However, deep learning algorithms can be overkill for less complex problems because they require access to a vast amount of data to be effective. For instance, ImageNet, the common benchmark for training deep learning models for comprehensive image recognition, has access to over 14 million images.
If the data is too simple or there isn’t enough if it, it is very easy for a deep learning model to become overfitted and fail to generalize well to new data. As a result, deep learning models don’t perform as well as other techniques like boosted decision trees or linear models for most practical business problems when there are smaller datasets and fewer features, like understanding customer churn or detecting fraudulent transactions. In certain cases like multiclass classification, deep learning can work for smaller, structured datasets.
Deep Learning + DataRobot
DataRobot’s automated machine learning platform includes support for deep learning and neural networks with technologies like TensorFlow. Additionally, DataRobot employs several cutting-edge techniques that make deep learning more effective on smaller, less complex datasets. DataRobot develops dozens of different types of models, allowing businesses to compare deep learning with various others in order to decide for themselves which models perform the best for the specific problem at hand.