Mobile and embedded platforms are increasingly required to efficiently execute computationally demanding DNNs across heterogeneous processing elements. At runtime, the available hardware resources to DNNs can vary considerably due to other concurrently running applications. The performance requirements of the applications could also change under different scenarios. To achieve the desired performance, dynamic DNNs have been proposed in which the number of channels/layers can be scaled in real time to meet different requirements under varying resource constraints. However, the training process of such dynamic DNNs can be costly, since platform-aware models of different deployment scenarios must be retrained to become dynamic. This paper proposes Dynamic-OFA, a novel dynamic DNN approach for state-of-the-art platform-aware NAS models (i.e. Once-for-all network (OFA)). Dynamic-OFA pre-samples a family of sub-networks from a static OFA backbone model, and contains a runtime manager to choose different sub-networks under different runtime environments. As such, Dynamic-OFA does not need the traditional dynamic DNN training pipeline. Compared to the state-of-the-art, our experimental results using ImageNet on a Jetson Xavier NX show that the approach is up to 3.5x (CPU), 2.4x (GPU) faster for similar Top-1 accuracy, or 3.8% (CPU), 5.1% (GPU) higher accuracy at similar latency.

A number of optimization approaches have been proposed for optimizing nonconvex objectives (e.g. deep learning models), such as batch gradient descent, stochastic gradient descent and stochastic variance reduced gradient descent. Theory shows these optimization methods can converge by using an unbiased gradient estimator. However, in practice biased gradient estimation can allow more e cient convergence to the vicinity since an unbiased approach is computationally more expensive. To produce fast convergence there are two trade-o s of these optimization strategies which are between stochastic/batch, and between biased/unbiased. This paper proposes an integrated approach which can control the nature of the stochastic element in the optimizer and can balance the trade-o of estimator between the biased and unbiased by using a hyper-parameter. It is shown theoretically and experimentally that this hyperparameter can be configured to provide an e ective balance to improve the convergence rate.

Recently, there has been growing interest in improving the efficiency and accuracy of the Indoor Positioning System (IPS). The Received Signal Strength- (RSS-) based fingerprinting technique is essential for indoor localization. However, it is challenging to estimate the indoor position based on RSS’s measurement under the complex indoor environment. This paper evaluates three machine learning approaches and Gaussian Process (GP) regression with three different kernels to get the best indoor positioning model. The hyperparameter tuning technique is used to select the optimum parameter set for each model. Experiments are carried out with RSS data from seven access points (AP). Results show that GP with a rational quadratic kernel and eXtreme gradient tree boosting model has the best positioning accuracy compared to other models. In contrast, the eXtreme gradient tree boosting model could achieve higher positioning accuracy with smaller training size and fewer access points.

No abstract