Search for: All records

Incremental learning is a challenging task in the field of machine learning, and it is a key step towards autonomous learning and adaptation. With the increasing attention on neuromorphic computing, there is an urgent need to investigate incremental learning techniques that can work in this paradigm to maintain energy efficiency while benefiting from flexibility and adaptability. In this paper, we present SEMINAR (sensitivity modulated importance networking and rehearsal), an incremental learning algorithm designed specifically for EMSTDP (Error Modulated Synaptic-Timing Dependent Plasticity), which performs supervised learning for multi-layer spiking neural networks (SNN) implemented on neuromorphic hardware, such as Loihi. SEMINAR uses critical synapse selection, differential learning rate and a replay buffer to enable the model to retain past knowledge while maintaining flexibility to learn new tasks. Our experimental results show that, when combined with the EMSTDP, SEMINAR outperforms different baseline incremental learning algorithms and gives more than 4% improvement on several widely used datasets such as Split-MNIST, Split-Fashion MNIST, Split-NMNIST and MSTAR. 

Modern SSDs achieve high throughput by utilizing multiple independent channels and chips in parallel. However, we find that excessive parallelism inadvertently amplifies the garbage collection (GC) overhead due to the larger unit of space reclamation. Based on this observation, we design PLAN, a novel SSD parallelism management and data placement scheme that allocates different levels of parallelism to different workloads with different needs to minimize the GC overhead. We demonstrate the effectiveness of PLAN by evaluating it against other state-of-the-art designs across various real-world workloads. PLAN reduces write amplification with comparable or better performance to the other designs that are always at full parallelism. 

A reliable command and control (C2) data link is required for unmanned aircraft systems (UAS) operations in order to monitor the status and support the control of UAS. A practical realization of the C2 communication and mission data links for commercial UAS operations is via LTE/5G networks. While the trajectory of each UAS directly determines the flight distance and mission cost in terms of energy dissipation, it also has a strong correlation to the quality of the communication link provided by a serving base station, where quality is defined as the achieved signal-to-interference-plus-noise ratio (SINR) required to maintain the control link of the UAS. Due to signal interference and the use of RF spectrum resources, the trajectory of a UAS not only determines the communication link quality it will encounter, but also influences the link quality of other UAS in its vicinity. Therefore, effective UAS traffic management must plan the trajectory for a group of UAS taking into account the impact to the interference levels of other base stations and UAS communication links. In this paper, an SINR Aware Predictive Planning (SAPP) framework is presented for trajectory planning of UAS leveraging 4G/5G communication networks in a simulated environment. The goal is to minimize flight distance while ensuring a minimum required link quality for C2 communications between UAS and base stations. The predictive control approach is proposed to address the challenges of the time varying SINR caused by the interference from other UAS’s communication. Experimental results show that the SAPP framework provides more than 3dB improvements on average for UAS communication parameters compared to traditional trajectory planning algorithms while still achieving shortest path trajectories and collision avoidance. 

Spiking neural networks(SNNs) have drawn broad research interests in recent years due to their high energy efficiency and biologically-plausibility. They have proven to be competitive in many machine learning tasks. Similar to all Artificial Neural Network(ANNs) machine learning models, the SNNs rely on the assumption that the training and testing data are drawn from the same distribution. As the environment changes gradually, the input distribution will shift over time, and the performance of SNNs turns out to be brittle. To this end, we propose a unified framework that can adapt nonstationary streaming data by exploiting unlabeled intermediate domain, and fits with the in-hardware SNN learning algorithm Error-modulated STDP. Specifically, we propose a unique self training framework to generate pseudo labels to retrain the model for intermediate and target domains. In addition, we develop an online-normalization method with an auxiliary neuron to normalize the output of the hidden layers. By combining the normalization with self-training, our approach gains average classification improvements over 10% on MNIST, NMINST, and two other datasets. 

We present IOTap, a tool that analyzes and profiles block I/O traces. IOTap computes the (dis)similarities among a set of workloads and sets a guideline for selecting a subset of traces for benchmarking. By doing so, we avoid experimentally running all workloads or, even worse, arbitrarily selecting a subset that skews the results.We demonstrate the usefulness of IOTap by comparing its results with experiments on real SSDs, achieving a high correlation of 0.92 for an NVMe SSD. 

Command and control (C2) data links over cellular networks is envisioned to be a reliable communications modality for various types of missions for Unmanned Aircraft System (UAS). The planning of UAS traffic and the provision of cellular communication resources are cross-coupled decisions that should be analyzed together to understand the quality of service such a modality can provide that meets business needs. The key to effective planning is the accurate estimation of communication link quality and the resource usage for a given air traffic requirement. In this work, a simulation and modelling framework is developed that integrates two open-source simulation platforms, Repast Simphony and ns-3, to generate UAS missions over different geographical areas and simulates the provision of 4G/5G cellular network connectivity to support their C2 and mission data links. To the best of our knowledge, this is the first simulator that co-simulates air traffic and cellular network communications for UAS while leveraging standardized 3GPP propagation models and incorporating detailed management of communication channels (i.e., resource blocks) at the cellular base station level. Three experiments were executed to demonstrate how the integrated simulation platform can be used to provide guidelines in communication resource allocation, air traffic management, and mission safety management in beyond visual line of sight (BVLOS) operations. 

The explosion of “big data” applications imposes severe challenges of speed and scalability on traditional computer systems. As the performance of traditional Von Neumann machines is greatly hindered by the increasing performance gap between CPU and memory (“known as the memory wall”), neuromorphic computing systems have gained considerable attention. The biology-plausible computing paradigm carries out computing by emulating the charging/discharging process of neuron and synapse potential. The unique spike domain information encoding enables asynchronous event driven computation and communication, and hence has the potential for very high energy efficiency. This survey reviews computing models and hardware platforms of existing neuromorphic computing systems. Neuron and synapse models are first introduced, followed by the discussion on how they will affect hardware design. Case studies of several representative hardware platforms, including their architecture and software ecosystems, are further presented. Lastly we present several future research directions. 

Spiking neural networks(SNNs) have drawn broad research interests in recent years due to their high energy efficiency and biologically-plausibility. They have proven to be competitive in many machine learning tasks. Similar to all Artificial Neural Network(ANNs) machine learning models, the SNNs rely on the assumption that the training and testing data are drawn from the same distribution. As the environment changes gradually, the input distribution will shift over time, and the performance of SNNs turns out to be brittle. To this end, we propose a unified framework that can adapt non-stationary streaming data by exploiting unlabeled intermediate domain, and fits with the in-hardware SNN learning algorithm Error-modulated STDP. Specifically, we propose a unique self-training framework to generate pseudo labels to retrain the model for intermediate and target domains. In addition, we develop an online-normalization method with an auxiliary neuron to normalize the output of the hidden layers. By combining the normalization with self-training, our approach gains average classification improvements over 10% on MNIST, NMINST, and two other datasets.