Search

In integrated circuit design, the analysis of wafer map patterns is critical to improve yield and detect manufacturing issues. We develop Wafer2Spike, an architecture for wafer map pattern classification using a spiking neural network (SNN), and demonstrate that a well-trained SNN achieves superior performance compared to deep neural network-based solutions. Wafer2Spike achieves an average classification accuracy of 98\% on the WM-811k wafer benchmark dataset. It is also superior to existing approaches for classifying defect patterns that are underrepresented in the original dataset. Wafer2Spike achieves this improved precision with great computational efficiency.

We introduce QUANTISENC, a fully configurable open-source software-defined digital quantized spiking neural core architecture to advance research in neuromorphic computing. QUANTISENC is designed hierarchically using a bottom-up methodology with multiple neurons in each layer and multiple layers in each core. The number of layers and neurons per layer can be configured via software in a top-down methodology to generate the hardware for a target spiking neural network (SNN) model. QUANTISENC uses leaky integrate and fire neurons (LIF) and current-based excitatory and inhibitory synapses (CUBA). The nonlinear dynamics of a neuron can be configured at run-time via programming its internal control registers. Each neuron performs signed fixed-point arithmetic with user-defined quantization and decimal precision. QUANTISENC supports all-to-all, one-to-one, and Gaussian connections between layers. Its hardware-software interface is integrated with a PyTorch-based SNN simulator. This integration allows to define and train an SNN model in PyTorch and evaluate the hardware performance (e.g., area, power, latency, and throughput) through FPGA prototyping and ASIC design. The hardware-software interface also takes advantage of the layer-based architecture and distributed memory organization of QUANTISENC to enable pipelining by overlapping computations on streaming data. Overall, the proposed software-defined hardware design methodology offers flexibility similar to that of high-level synthesis (HLS), but provides better hardware performance with zero hardware development effort. We evaluate QUANTISENC using three spiking datasets and show its superior performance against state-of the-art designs.

Sixth-generation (6G) communication systems are poised to accommodate high data-rate wireless communication services in highly dynamic channels, with applications including high-speed trains, unmanned aerial vehicles, and intelligent transportation systems. Orthogonal frequency-division multiplexing (OFDM) modulation suffers from performance degradation in such high-mobility applications due to high Doppler spread in the channel. The recently proposed Orthogonal Time Frequency Space (OTFS) modulation scheme outperforms OFDM in terms of supporting a higher transmitter (Tx) and receiver (Rx) user velocity. Additionally, the highly-dynamic time-frequency (TF) channel has little effect on OTFS modulated signals, which enables the realization of low-complexity pre-processing architectures for implementing massive-multiple input multiple outputs (MIMO) based OTFS systems. However, while OTFS has received attention in the literature from a theory and simulation perspective, there has been comparatively little work on real-time FPGA implementation of OTFS waveforms. Thus, in this paper, we first present a mathematical overview of OTFS modulation and then describe an FPGA implementation of OTFS implementation on hardware. Power, area, and timing analysis of the implemented design on a Zynq UltraScale+ RFSoC FPGA are provided for benchmarking purposes., Comment: Accepted at IEEE Future Networks, 2023

Neuromorphic Systems-on-Chip (NSoCs) are becoming heterogeneous by integrating general-purpose processors (GPPs) and neural processing units (NPUs) on the same SoC. For embedded systems, an NSoC may need to execute user applications built using a variety of machine learning models. We propose a real-time scheduler, called PRISM, which can schedule machine learning models on a heterogeneous NSoC either individually or concurrently to improve their system performance. PRISM consists of the following four key steps. First, it constructs an interprocessor communication (IPC) graph of a machine learning model from a mapping and a self-timed schedule. Second, it creates a transaction order for the communication actors and embeds this order into the IPC graph. Third, it schedules the graph on an NSoC by overlapping communication with the computation. Finally, it uses a Hill Climbing heuristic to explore the design space of mapping operations on GPPs and NPUs to improve the performance. Unlike existing schedulers which use only the NPUs of an NSoC, PRISM improves performance by enabling batch, pipeline, and operation parallelism via exploiting a platform's heterogeneity. For use-cases with concurrent applications, PRISM uses a heuristic resource sharing strategy and a non-preemptive scheduling to reduce the expected wait time before concurrent operations can be scheduled on contending resources. Our extensive evaluations with 20 machine learning workloads show that PRISM significantly improves the performance per watt for both individual applications and use-cases when compared to state-of-the-art schedulers., Comment: To appear at MEMOCODE 2022

Guillain-Barre syndrome is a rare neurological condition in which the human immune system attacks the peripheral nervous system. A peripheral nervous system appears as a diffusively connected system of mathematical models of neuron models, and the system's period becomes shorter than the periods of each neural circuit. The stimuli in the conduction path that will address the myelin sheath that has lost its function are received by the axons and are conveyed externally to the target organ, aiming to solve the problem of decreased nerve conduction. In the NEURON simulation environment, one can create a neuron model and define biophysical events that take place within the system for study. In this environment, signal transmission between cells and dendrites is obtained graphically. The simulated potassium and sodium conductance are replicated adequately, and the electronic action potentials are quite comparable to those measured experimentally. In this work, we propose an analog and digital coupled neuron model comprising individual excitatory and inhibitory neural circuit blocks for a low-cost and energy-efficient system. Compared to digital design, our analog design performs in lower frequency but gives a 32.3\% decreased energy efficiency. Thus, the resulting coupled analog hardware neuron model can be a proposed model for the simulation of reduced nerve conduction. As a result, the analog coupled neuron, (even with its greater design complexity) serious contender for the future development of a wearable sensor device that could help with Guillain-Barre syndrome and other neurologic diseases.

Reconstructing perceived images from human brain activity monitored by functional magnetic resonance imaging (fMRI) is hard, especially for natural images. Existing methods often result in blurry and unintelligible reconstructions with low fidelity. In this study, we present a novel approach for enhanced image reconstruction, in which existing methods for object decoding and image reconstruction are merged together. This is achieved by conditioning the reconstructed image to its decoded image category using a class-conditional generative adversarial network and neural style transfer. The results indicate that our approach improves the semantic similarity of the reconstructed images and can be used as a general framework for enhanced image reconstruction., Comment: Accepted at 2022 International Joint Conference on Neural Networks

Feedback-driven recurrent spiking neural networks (RSNNs) are powerful computational models that can mimic dynamical systems. However, the presence of a feedback loop from the readout to the recurrent layer de-stabilizes the learning mechanism and prevents it from converging. Here, we propose a supervised training procedure for RSNNs, where a second network is introduced only during the training, to provide hint for the target dynamics. The proposed training procedure consists of generating targets for both recurrent and readout layers (i.e., for a full RSNN system). It uses the recursive least square-based First-Order and Reduced Control Error (FORCE) algorithm to fit the activity of each layer to its target. The proposed full-FORCE training procedure reduces the amount of modifications needed to keep the error between the output and target close to zero. These modifications control the feedback loop, which causes the training to converge. We demonstrate the improved performance and noise robustness of the proposed full-FORCE training procedure to model 8 dynamical systems using RSNNs with leaky integrate and fire (LIF) neurons and rate coding. For energy-efficient hardware implementation, an alternative time-to-first-spike (TTFS) coding is implemented for the full- FORCE training procedure. Compared to rate coding, full-FORCE with TTFS coding generates fewer spikes and facilitates faster convergence to the target dynamics., Comment: Accepted at IJCNN 2022

We propose a design methodology to facilitate fault tolerance of deep learning models. First, we implement a many-core fault-tolerant neuromorphic hardware design, where neuron and synapse circuitries in each neuromorphic core are enclosed with astrocyte circuitries, the star-shaped glial cells of the brain that facilitate self-repair by restoring the spike firing frequency of a failed neuron using a closed-loop retrograde feedback signal. Next, we introduce astrocytes in a deep learning model to achieve the required degree of tolerance to hardware faults. Finally, we use a system software to partition the astrocyte-enabled model into clusters and implement them on the proposed fault-tolerant neuromorphic design. We evaluate this design methodology using seven deep learning inference models and show that it is both area and power efficient., Comment: Accepted at ACM Computing Frontiers, 2022

Neuromorphic hardware platforms can significantly lower the energy overhead of a machine learning inference task. We present a design-technology tradeoff analysis to implement such inference tasks on the processing elements (PEs) of a Non- Volatile Memory (NVM)-based neuromorphic hardware. Through detailed circuit-level simulations at scaled process technology nodes, we show the negative impact of technology scaling on the information-processing latency, which impacts the quality-of-service (QoS) of an embedded ML system. At a finer granularity, the latency inside a PE depends on 1) the delay introduced by parasitic components on its current paths, and 2) the varying delay to sense different resistance states of its NVM cells. Based on these two observations, we make the following three contributions. First, on the technology front, we propose an optimization scheme where the NVM resistance state that takes the longest time to sense is set on current paths having the least delay, and vice versa, reducing the average PE latency, which improves the QoS. Second, on the architecture front, we introduce isolation transistors within each PE to partition it into regions that can be individually power-gated, reducing both latency and energy. Finally, on the system-software front, we propose a mechanism to leverage the proposed technological and architectural enhancements when implementing a machine-learning inference task on neuromorphic PEs of the hardware. Evaluations with a recent neuromorphic hardware architecture show that our proposed design-technology co-optimization approach improves both performance and energy efficiency of machine-learning inference tasks without incurring high cost-per-bit., Comment: Accepted for publication at ACM TECS

Precise monitoring of respiratory rate in premature infants is essential to initiate medical interventions as required. Wired technologies can be invasive and obtrusive to the patients. We propose a Deep Learning enabled wearable monitoring system for premature newborn infants, where respiratory cessation is predicted using signals that are collected wirelessly from a non-invasive wearable Bellypatch put on infant's body. We propose a five-stage design pipeline involving data collection and labeling, feature scaling, model selection with hyperparameter tuning, model training and validation, model testing and deployment. The model used is a 1-D Convolutional Neural Network (1DCNN) architecture with 1 convolutional layer, 1 pooling layer and 3 fully-connected layers, achieving 97.15% accuracy. To address energy limitations of wearable processing, several quantization techniques are explored and their performance and energy consumption are analyzed. We propose a novel Spiking-Neural-Network(SNN) based respiratory classification solution, which can be implemented on event-driven neuromorphic hardware. We propose an approach to convert the analog operations of our baseline 1DCNN to their spiking equivalent. We perform a design-space exploration using the parameters of the converted SNN to generate inference solutions having different accuracy and energy footprints. We select a solution that achieves 93.33% accuracy with 18 times lower energy compared with baseline 1DCNN model. Additionally the proposed SNN solution achieves similar accuracy but with 4 times less energy.

Recently, both industry and academia have proposed several different neuromorphic systems to execute machine learning applications that are designed using Spiking Neural Networks (SNNs). With the growing complexity on design and technology fronts, programming such systems to admit and execute a machine learning application is becoming increasingly challenging. Additionally, neuromorphic systems are required to guarantee real-time performance, consume lower energy, and provide tolerance to logic and memory failures. Consequently, there is a clear need for system software frameworks that can implement machine learning applications on current and emerging neuromorphic systems, and simultaneously address performance, energy, and reliability. Here, we provide a comprehensive overview of such frameworks proposed for both, platform-based design and hardware-software co-design. We highlight challenges and opportunities that the future holds in the area of system software technology for neuromorphic computing.