Deep Neural Networks (DNNs) have gained widespread popularity in different domain applications due to their dominant performance. Despite the prevalence of massively parallel multi-core processor architectures, adopting large DNN models in embedded systems remains challenging, as most embedded applications are designed with single-core processors in mind. This limits DNN adoption in embedded systems due to inefficient leveraging of model parallelization and workload partitioning. Prior solutions attempt to address these challenges using data and model parallelism. However, they lack in finding optimal DNN model partitions and distributing them efficiently to achieve improved performance.This paper proposes a DNN model parallelism framework to accelerate model training by finding the optimal number of model partitions and resource provisions. The proposed framework combines data and model parallelism techniques to optimize the parallel processing of DNNs for embedded applications. In addition, it implements the pipeline execution of the partitioned models and integrates a task controller to manage the computing resources. The experimental results for image object detection demonstrate the applicability of our proposed framework in estimating the latest execution time and reducing overall model training time by almost 44.87% compared to the baseline AlexNet convolutional neural network (CNN) model.

In this article, we present an edge-based architecture for enhancing the autonomous capabilities of resource-constrained aerial robots by enabling a remote nonlinear model predictive control scheme, which can be computationally heavy to run on the aerial robots' onboard processors. The nonlinear model predictive control is used to control the trajectory of an unmanned aerial vehicle while detecting, and preventing potential collisions. The proposed edge architecture enables trajectory recalculation for resource-constrained unmanned aerial vehicles in relatively real-time, which will allow them to have fully autonomous behaviors. The architecture is implemented with a remote Kubernetes cluster on the edge side, and it is evaluated on an unmanned aerial vehicle as our controllable robot, while the robotic operating system is used for managing the source codes, and overall communication. With the utilization of edge computing and the architecture presented in this work, we can overcome computational limitations, that resource-constrained robots have, and provide or improve features that are essential for autonomous missions. At the same time, we can minimize the relative travel time delays for time-critical missions over the edge, in comparison to the cloud. We investigate the validity of this hypothesis by evaluating the system's behavior through a series of experiments by utilizing either the unmanned aerial vehicle or the edge resources for the collision avoidance mission.

Finding a novel drug is a very long and complex procedure. Using computer simulations, it is possible to accelerate the preliminary phases by performing a virtual screening that filters a large set of drug candidates to a manageable number. This paper presents the implementations and comparative analysis of two GPU-optimized implementations of a virtual screening algorithm targeting novel GPU architectures. This work focuses on the analysis of parallel computation patterns and their mapping onto the target architecture. The first method adopts a traditional approach that spreads the computation for a single molecule across the entire GPU. The second uses a novel batched approach that exploits the parallel architecture of the GPU to evaluate more molecules in parallel. Experimental results showed a different behavior depending on the size of the database to be screened, either reaching a performance plateau sooner or having a more extended initial transient period to achieve a higher throughput (up to 5x), which is more suitable for extreme-scale virtual screening campaigns.