{ The introductory machine learning MNIST digit data-set contains images of hand-written digits from 0 - 9. A classi er model would be required to study an input image and determine its corresponding digit (0, 1, 2 ..., 8 or 9). { Symptoms of patients are used to determine their corresponding disease. A classi er model could be built to study symptoms and predict diseases automatically.
Deep learning methods for classi er development have been found wanting when translated to real-world applications. The most notable drawbacks of deep learning is the cost associated with gathering annotated samples and the assumption that all is known about the domain. Within this work we study and alter deep learning techniques to develop operational-grade classi er models whilst addressing these two limitations.
For classi cation, it is assumed that each data sample encountered by the model will belong to some category/class in the domain which can be determined through studying of the input samples. Some classi cation examples are:
In the developing world where the majority of the population cannot a ord doctor visits due to lack of doctors and/or money, having a classi cation system capable of quick free diagnoses is of immense bene t. This example is one of many which shows the potential of the 4th industrial revolution.
Deep learning has in recent years seen growing success in training classi er models. Typically deep models are developed/trained using two learning regimes, supervised or semi-supervised learning.
Supervised learning requires that each input sample used to train the model
has to be annotated/labelled to indicate which class it belongs to. Models then
study (many) labelled samples and learn the di erences between the various
classes in the domain. Given that the model trained well, new never-before seen
input data can be accurately classi ed during the operation/testing phase.
Supervised learning has shown high accuracy scores with deep neural networks
given that a large number of annotated samples are provided [
Emile Engelbrecht and Johan du Preez annotating samples within applications, however, is expensive which lead to the development of the second learning regime, namely semi-supervised learning.
Semi-supervised learning research has achieved similar accuracy results to
supervised learning while requiring much less labelled data [
Within real-world application domains there might be (many) classes which are not known to designers. For the symptoms/disease example above, unknown classes might be the symptom-disease correlations yet to be discovered by doctors. In such cases there is no certainty as to whether an unlabelled sample corresponds to one of the classes known to the system or not. This violates the assumption of semi-supervised training, causing it to break down. It still, however, remains fundamental for operative classi ers to classify over all classes in the domain, known or unknown, as samples from all classes will be encountered.
To the best of our knowledge, no work has addressed a semi-supervised learning scenario where unlabelled samples might also belong to categories outside of those known to the classi cation system. We therefore rst formally de ne this setting, which we coin a quasi-open set, after which we propose a learning regime to handle quasi-open sets. This learning regime is called quasi-open set semi-supervised training and requires known classes be correctly classi ed whilst simultaneously requiring unknown classes (only seen in the unlabelled data) be classi ed as 'other'.
Training under quasi-closed semi-supervised learning develops a model capable of classifying over all classes in the domain even though each class might not be explicitly expressed to the system. Models are therefore able to train using vast un ltered and unrestricted unlabelled sets (as would be available in application) which was previously un-attainable using deep neural networks.
Our proposed method for quasi-closed semi-supervised learning uses
generative adversarial networks (GANs) in a similar fashion to general semi-supervised
learning using GANs [
With this new learning regime and our proposed method, we are able to build operational classi ers (capable of classifying over all classes in the domain) without requiring each individual class be made known to the system. This approach greatly extends the range of automated classi cation tasks that can be addressed in a cost e ective manner.
Quasi-closed semi-supervised learning