Deep Convolutional Neural Network Architecture

These explanatory diagrams are used to concisely describe the architecture of the client’s proprietary Deep Convolutional Neural Network, which forms an integral part of their diagnostics and drug development process. Using fluorescent microscopy images as input, the network quantifies the oncogenicity levels of synthesized mutations and classifies them as driver or passenger mutations, thus helping to identify potential molecular targets or assess the efficacy of specific inhibitors.

The diagrams address an expert audience and are used in conferences, publications and marketing materials such as investor decks, to efficiently communicate the machine learning component of the client’s technology.

The network’s multiple layers, components and iterations are mapped out in a cohesive visual presentation, using various shorthands clear to those already familiar with the basics of the subject. This makes is possible to communicate both the general architecture and relevant details in a single visual, whilst avoiding information overload. In this way, the presenter can refer to and explain each part of the network without the audience losing sight of the big picture.