Charles Martin

Charles Martin, Henry Gardner, Ben Swift

We present and evaluate a novel interface for tracking ensemble performances on touch-screens. The system uses a Random Forest classifier to extract touch-screen gestures and transition matrix statistics. It analyses the resulting gesture-state sequences across an ensemble of performers. A series of specially designed iPad apps respond to this real-time analysis of free-form gestural performances with calculated modifications to their musical interfaces. We describe our system and evaluate it through cross-validation and profiling as well as concert experience.

Charles Martin

This paper describes the development of an Apple iPhone based mobile computer system for vibraphone and its use in a series of the author's performance projects in 2011 and 2012. This artistic research was motivated by a desire to develop an alternative to laptop computers for the author's existing percussion and computer performance practice. The aims were to develop a light, compact and flexible system using mobile devices that would allow computer music to infiltrate solo and ensemble performance situations where it is difficult to use a laptop computer. The project began with a system that brought computer elements to Nordlig Vinter, a suite of percussion duos, using an iPhone, RjDj, Pure Data and a home-made pickup system. This process was documented with video recordings and analysed using ethnographic methods. The mobile computer music setup proved to be elegant and convenient in performance situations with very little time and space to set up, as well as in performance classes and workshops. The simple mobile system encouraged experimentation and the platforms used enabled sharing with a wider audience.

Charles Martin, Chi-Hsia Lai

This paper describes Strike on Stage, an interface and corresponding audio-visual performance work developed and performed in 2010 by percussionists and media artists Chi-Hsia Lai and Charles Martin. The concept of Strike on Stage is to integrate computer visuals and sound into an improvised percussion performance. A large projection surface is positioned directly behind the performers, while a computer vision system tracks their movements. The setup allows computer visualisation and sonification to be directly responsive and unified with the performers' gestures.

Charles Martin, Benjamin Forster, Hanna Cormick

In 2009 the cross artform group, Last Man to Die, presented a series of performances using new interfaces and networked performance to integrate the three artforms of its members (actor, Hanna Cormick, visual artist, Benjamin Forster and percussionist, Charles Martin). This paper explains our artistic motivations and design for a computer vision surface and networked heartbeat sensor as well as the experience of mounting our first major work, Vital LMTD.

Phillip Pope, Soheil Kolouri, Mohammad Rostrami et al.

Functional groups (FGs) are molecular substructures that are served as a foundation for analyzing and predicting chemical properties of molecules. Automatic discovery of FGs will impact various fields of research, including medicinal chemistry and material sciences, by reducing the amount of lab experiments required for discovery or synthesis of new molecules. In this paper, we investigate methods based on graph convolutional neural networks (GCNNs) for localizing FGs that contribute to specific chemical properties of interest. In our framework, molecules are modeled as undirected relational graphs with atoms as nodes and bonds as edges. Using this relational graph structure, we trained GCNNs in a supervised way on experimentally-validated molecular training sets to predict specific chemical properties, e.g., toxicity. Upon learning a GCNN, we analyzed its activation patterns to automatically identify FGs using four different explainability methods that we have developed: gradient-based saliency maps, Class Activation Mapping (CAM), gradient-weighted CAM (Grad-CAM), and Excitation Back-Propagation. Although these methods are originally derived for convolutional neural networks (CNNs), we adapt them to develop the corresponding suitable versions for GCNNs. We evaluated the contrastive power of these methods with respect to the specificity of the identified molecular substructures and their relevance for chemical functions. Grad-CAM had the highest contrastive power and generated qualitatively the best FGs. This work paves the way for automatic analysis and design of new molecules.