Corey Lammie

Manuel Le Gallo, Corey Lammie, Julian Buechel et al.

Analog In-Memory Computing (AIMC) is a promising approach to reduce the latency and energy consumption of Deep Neural Network (DNN) inference and training. However, the noisy and non-linear device characteristics, and the non-ideal peripheral circuitry in AIMC chips, require adapting DNNs to be deployed on such hardware to achieve equivalent accuracy to digital computing. In this tutorial, we provide a deep dive into how such adaptations can be achieved and evaluated using the recently released IBM Analog Hardware Acceleration Kit (AIHWKit), freely available at https://github.com/IBM/aihwkit. The AIHWKit is a Python library that simulates inference and training of DNNs using AIMC. We present an in-depth description of the AIHWKit design, functionality, and best practices to properly perform inference and training. We also present an overview of the Analog AI Cloud Composer, a platform that provides the benefits of using the AIHWKit simulation in a fully managed cloud setting along with physical AIMC hardware access, freely available at https://aihw-composer.draco.res.ibm.com. Finally, we show examples on how users can expand and customize AIHWKit for their own needs. This tutorial is accompanied by comprehensive Jupyter Notebook code examples that can be run using AIHWKit, which can be downloaded from https://github.com/IBM/aihwkit/tree/master/notebooks/tutorial.

Corey Lammie

Analog In-Memory Computing (AIMC) accelerators execute matrix-vector multiplications directly within memory arrays, reducing data movement and improving DNN inference efficiency. Their limited effective precision motivates heterogeneous architectures that combine analog compute tiles with digital processing units. This letter classifies existing methods for partitioning DNN workloads across these resources by mapping granularity, optimization strategy, and model support, and distills them into a unified four-stage workflow. To demonstrate the workflow on a model class not yet addressed by existing methods, we apply its first two stages to GPT-2, producing the first AIMC-specific precision sensitivity profile for a decoder-only transformer. Sensitivity is dominated by 4 of 49 projections, with the first decoder block's attention output dominating by an order of magnitude. This suggests that projection-level mapping and selective digital execution of early-block and output-facing projections are important for reliable decoder-transformer deployment on AIMC hardware.

Chenqi Li, Corey Lammie, Xuening Dong et al.

During the past two decades, epileptic seizure detection and prediction algorithms have evolved rapidly. However, despite significant performance improvements, their hardware implementation using conventional technologies, such as Complementary Metal-Oxide-Semiconductor (CMOS), in power and area-constrained settings remains a challenging task; especially when many recording channels are used. In this paper, we propose a novel low-latency parallel Convolutional Neural Network (CNN) architecture that has between 2-2,800x fewer network parameters compared to SOTA CNN architectures and achieves 5-fold cross validation accuracy of 99.84% for epileptic seizure detection, and 99.01% and 97.54% for epileptic seizure prediction, when evaluated using the University of Bonn Electroencephalogram (EEG), CHB-MIT and SWEC-ETHZ seizure datasets, respectively. We subsequently implement our network onto analog crossbar arrays comprising Resistive Random-Access Memory (RRAM) devices, and provide a comprehensive benchmark by simulating, laying out, and determining hardware requirements of the CNN component of our system. To the best of our knowledge, we are the first to parallelize the execution of convolution layer kernels on separate analog crossbars to enable 2 orders of magnitude reduction in latency compared to SOTA hybrid Memristive-CMOS DL accelerators. Furthermore, we investigate the effects of non-idealities on our system and investigate Quantization Aware Training (QAT) to mitigate the performance degradation due to low ADC/DAC resolution. Finally, we propose a stuck weight offsetting methodology to mitigate performance degradation due to stuck RON/ROFF memristor weights, recovering up to 32% accuracy, without requiring retraining. The CNN component of our platform is estimated to consume approximately 2.791W of power while occupying an area of 31.255mm$^2$ in a 22nm FDSOI CMOS process.

Tim Zhang, Corey Lammie, Mostafa Rahimi Azghadi et al.

Spike sorting algorithms are used to separate extracellular recordings of neuronal populations into single-unit spike activities. The development of customized hardware implementing spike sorting algorithms is burgeoning. However, there is a lack of a systematic approach and a set of standardized evaluation criteria to facilitate direct comparison of both software and hardware implementations. In this paper, we formalize a set of standardized criteria and a publicly available synthetic dataset entitled Synthetic Simulations Of Extracellular Recordings (SSOER), which was constructed by aggregating existing synthetic datasets with varying Signal-To-Noise Ratios (SNRs). Furthermore, we present a benchmark for future comparison, and use our criteria to evaluate a simulated Resistive Random-Access Memory (RRAM) In-Memory Computing (IMC) system using the Discrete Wavelet Transform (DWT) for feature extraction. Our system consumes approximately (per channel) 10.72mW and occupies an area of 0.66mm$^2$ in a 22nm FDSOI Complementary Metal-Oxide-Semiconductor (CMOS) process.

Hadjer Benmeziane, Corey Lammie, Irem Boybat et al.

The advancement of Deep Learning (DL) is driven by efficient Deep Neural Network (DNN) design and new hardware accelerators. Current DNN design is primarily tailored for general-purpose use and deployment on commercially viable platforms. Inference at the edge requires low latency, compact and power-efficient models, and must be cost-effective. Digital processors based on typical von Neumann architectures are not conducive to edge AI given the large amounts of required data movement in and out of memory. Conversely, analog/mixed signal in-memory computing hardware accelerators can easily transcend the memory wall of von Neuman architectures when accelerating inference workloads. They offer increased area and power efficiency, which are paramount in edge resource-constrained environments. In this paper, we propose AnalogNAS, a framework for automated DNN design targeting deployment on analog In-Memory Computing (IMC) inference accelerators. We conduct extensive hardware simulations to demonstrate the performance of AnalogNAS on State-Of-The-Art (SOTA) models in terms of accuracy and deployment efficiency on various Tiny Machine Learning (TinyML) tasks. We also present experimental results that show AnalogNAS models achieving higher accuracy than SOTA models when implemented on a 64-core IMC chip based on Phase Change Memory (PCM). The AnalogNAS search code is released: https://github.com/IBM/analog-nas

Jason K. Eshraghian, Corey Lammie, Mostafa Rahimi Azghadi et al.

Spiking and Quantized Neural Networks (NNs) are becoming exceedingly important for hyper-efficient implementations of Deep Learning (DL) algorithms. However, these networks face challenges when trained using error backpropagation, due to the absence of gradient signals when applying hard thresholds. The broadly accepted trick to overcoming this is through the use of biased gradient estimators: surrogate gradients which approximate thresholding in Spiking Neural Networks (SNNs), and Straight-Through Estimators (STEs), which completely bypass thresholding in Quantized Neural Networks (QNNs). While noisy gradient feedback has enabled reasonable performance on simple supervised learning tasks, it is thought that such noise increases the difficulty of finding optima in loss landscapes, especially during the later stages of optimization. By periodically boosting the Learning Rate (LR) during training, we expect the network can navigate unexplored solution spaces that would otherwise be difficult to reach due to local minima, barriers, or flat surfaces. This paper presents a systematic evaluation of a cosine-annealed LR schedule coupled with weight-independent adaptive moment estimation as applied to Quantized SNNs (QSNNs). We provide a rigorous empirical evaluation of this technique on high precision and 4-bit quantized SNNs across three datasets, demonstrating (close to) state-of-the-art performance on the more complex datasets. Our source code is available at this link: https://github.com/jeshraghian/QSNNs.

Elena Ferro, Athanasios Vasilopoulos, Corey Lammie et al.

Analog In-Memory Computing (AIMC) is an emerging technology for fast and energy-efficient Deep Learning (DL) inference. However, a certain amount of digital post-processing is required to deal with circuit mismatches and non-idealities associated with the memory devices. Efficient near-memory digital logic is critical to retain the high area/energy efficiency and low latency of AIMC. Existing systems adopt Floating Point 16 (FP16) arithmetic with limited parallelization capability and high latency. To overcome these limitations, we propose a Near-Memory digital Processing Unit (NMPU) based on fixed-point arithmetic. It achieves competitive accuracy and higher computing throughput than previous approaches while minimizing the area overhead. Moreover, the NMPU supports standard DL activation steps, such as ReLU and Batch Normalization. We perform a physical implementation of the NMPU design in a 14 nm CMOS technology and provide detailed performance, power, and area assessments. We validate the efficacy of the NMPU by using data from an AIMC chip and demonstrate that a simulated AIMC system with the proposed NMPU outperforms existing FP16-based implementations, providing 139$\times$ speed-up, 7.8$\times$ smaller area, and a competitive power consumption. Additionally, our approach achieves an inference accuracy of 86.65 %/65.06 %, with an accuracy drop of just 0.12 %/0.4 % compared to the FP16 baseline when benchmarked with ResNet9/ResNet32 networks trained on the CIFAR10/CIFAR100 datasets, respectively.

Chen Li, Elena Ferro, Corey Lammie et al.

Analog In-Memory Computing (AIMC) offers a promising solution to the von Neumann bottleneck. However, deploying transformer models on AIMC remains challenging due to their inherent need for flexibility and adaptability across diverse tasks. For the benefits of AIMC to be fully realized, weights of static vector-matrix multiplications must be mapped and programmed to analog devices in a weight-stationary manner. This poses two challenges for adapting a base network to hardware and downstream tasks: (i) conventional analog hardware-aware (AHWA) training requires retraining the entire model, and (ii) reprogramming analog devices is both time- and energy-intensive. To address these issues, we propose Analog Hardware-Aware Low-Rank Adaptation (AHWA-LoRA) training, a novel approach for efficiently adapting transformers to AIMC hardware. AHWA-LoRA training keeps the analog weights fixed as meta-weights and introduces lightweight external LoRA modules for both hardware and task adaptation. We validate AHWA-LoRA training on SQuAD v1.1 and the GLUE benchmark, demonstrate its scalability to larger models, and show its effectiveness in instruction tuning and reinforcement learning. We further evaluate a practical deployment scenario that balances AIMC tile latency with digital LoRA processing using optimized pipeline strategies, with RISC-V-based programmable multi-core accelerators. This hybrid architecture achieves efficient transformer inference with only a 4% per-layer overhead compared to a fully AIMC implementation.

Julian Büchel, Giacomo Camposampiero, Athanasios Vasilopoulos et al.

Kernel functions are vital ingredients of several machine learning algorithms, but often incur significant memory and computational costs. We introduce an approach to kernel approximation in machine learning algorithms suitable for mixed-signal Analog In-Memory Computing (AIMC) architectures. Analog In-Memory Kernel Approximation addresses the performance bottlenecks of conventional kernel-based methods by executing most operations in approximate kernel methods directly in memory. The IBM HERMES Project Chip, a state-of-the-art phase-change memory based AIMC chip, is utilized for the hardware demonstration of kernel approximation. Experimental results show that our method maintains high accuracy, with less than a 1% drop in kernel-based ridge classification benchmarks and within 1% accuracy on the Long Range Arena benchmark for kernelized attention in Transformer neural networks. Compared to traditional digital accelerators, our approach is estimated to deliver superior energy efficiency and lower power consumption. These findings highlight the potential of heterogeneous AIMC architectures to enhance the efficiency and scalability of machine learning applications.

Corey Lammie, Jason K. Eshraghian, Chenqi Li et al.

The impact of device and circuit-level effects in mixed-signal Resistive Random Access Memory (RRAM) accelerators typically manifest as performance degradation of Deep Learning (DL) algorithms, but the degree of impact varies based on algorithmic features. These include network architecture, capacity, weight distribution, and the type of inter-layer connections. Techniques are continuously emerging to efficiently train sparse neural networks, which may have activation sparsity, quantization, and memristive noise. In this paper, we present an extended Design Space Exploration (DSE) methodology to quantify the benefits and limitations of dense and sparse mapping schemes for a variety of network architectures. While sparsity of connectivity promotes less power consumption and is often optimized for extracting localized features, its performance on tiled RRAM arrays may be more susceptible to noise due to under-parameterization, when compared to dense mapping schemes. Moreover, we present a case study quantifying and formalizing the trade-offs of typical non-idealities introduced into 1-Transistor-1-Resistor (1T1R) tiled memristive architectures and the size of modular crossbar tiles using the CIFAR-10 dataset.

Alzayat Saleh, Issam H. Laradji, Corey Lammie et al.

Health professionals extensively use Two- Dimensional (2D) Ultrasound (US) videos and images to visualize and measure internal organs for various purposes including evaluation of muscle architectural changes. US images can be used to measure abdominal muscles dimensions for the diagnosis and creation of customized treatment plans for patients with Low Back Pain (LBP), however, they are difficult to interpret. Due to high variability, skilled professionals with specialized training are required to take measurements to avoid low intra-observer reliability. This variability stems from the challenging nature of accurately finding the correct spatial location of measurement endpoints in abdominal US images. In this paper, we use a Deep Learning (DL) approach to automate the measurement of the abdominal muscle thickness in 2D US images. By treating the problem as a localization task, we develop a modified Fully Convolutional Network (FCN) architecture to generate blobs of coordinate locations of measurement endpoints, similar to what a human operator does. We demonstrate that using the TrA400 US image dataset, our network achieves a Mean Absolute Error (MAE) of 0.3125 on the test set, which almost matches the performance of skilled ultrasound technicians. Our approach can facilitate next steps for automating the process of measurements in 2D US images, while reducing inter-observer as well as intra-observer variability for more effective clinical outcomes.

Corey Lammie, Jason K. Eshraghian, Wei D. Lu et al.

Stochastic Computing (SC) is a computing paradigm that allows for the low-cost and low-power computation of various arithmetic operations using stochastic bit streams and digital logic. In contrast to conventional representation schemes used within the binary domain, the sequence of bit streams in the stochastic domain is inconsequential, and computation is usually non-deterministic. In this brief, we exploit the stochasticity during switching of probabilistic Conductive Bridging RAM (CBRAM) devices to efficiently generate stochastic bit streams in order to perform Deep Learning (DL) parameter optimization, reducing the size of Multiply and Accumulate (MAC) units by 5 orders of magnitude. We demonstrate that in using a 40-nm Complementary Metal Oxide Semiconductor (CMOS) process our scalable architecture occupies 1.55mm$^2$ and consumes approximately 167$μ$W when optimizing parameters of a Convolutional Neural Network (CNN) while it is being trained for a character recognition task, observing no notable reduction in accuracy post-training.

Corey Lammie, Wei Xiang, Mostafa Rahimi Azghadi

The unpredictability of seizures continues to distress many people with drug-resistant epilepsy. On account of recent technological advances, considerable efforts have been made using different hardware technologies to realize smart devices for the real-time detection and prediction of seizures. In this paper, we investigate the feasibility of using Memristive Deep Learning Systems (MDLSs) to perform real-time epileptic seizure prediction on the edge. Using the MemTorch simulation framework and the Children's Hospital Boston (CHB)-Massachusetts Institute of Technology (MIT) dataset we determine the performance of various simulated MDLS configurations. An average sensitivity of 77.4% and a Area Under the Receiver Operating Characteristic Curve (AUROC) of 0.85 are reported for the optimal configuration that can process Electroencephalogram (EEG) spectrograms with 7,680 samples in 1.408ms while consuming 0.0133W and occupying an area of 0.1269mm$^2$ in a 65nm Complementary Metal-Oxide-Semiconductor (CMOS) process.

Mostafa Rahimi Azghadi, Corey Lammie, Jason K. Eshraghian et al.

The advent of dedicated Deep Learning (DL) accelerators and neuromorphic processors has brought on new opportunities for applying both Deep and Spiking Neural Network (SNN) algorithms to healthcare and biomedical applications at the edge. This can facilitate the advancement of medical Internet of Things (IoT) systems and Point of Care (PoC) devices. In this paper, we provide a tutorial describing how various technologies including emerging memristive devices, Field Programmable Gate Arrays (FPGAs), and Complementary Metal Oxide Semiconductor (CMOS) can be used to develop efficient DL accelerators to solve a wide variety of diagnostic, pattern recognition, and signal processing problems in healthcare. Furthermore, we explore how spiking neuromorphic processors can complement their DL counterparts for processing biomedical signals. The tutorial is augmented with case studies of the vast literature on neural network and neuromorphic hardware as applied to the healthcare domain. We benchmark various hardware platforms by performing a sensor fusion signal processing task combining electromyography (EMG) signals with computer vision. Comparisons are made between dedicated neuromorphic processors and embedded AI accelerators in terms of inference latency and energy. Finally, we provide our analysis of the field and share a perspective on the advantages, disadvantages, challenges, and opportunities that various accelerators and neuromorphic processors introduce to healthcare and biomedical domains.

Corey Lammie, Wei Xiang, Mostafa Rahimi Azghadi

While hardware implementations of inference routines for Binarized Neural Networks (BNNs) are plentiful, current realizations of efficient BNN hardware training accelerators, suitable for Internet of Things (IoT) edge devices, leave much to be desired. Conventional BNN hardware training accelerators perform forward and backward propagations with parameters adopting binary representations, and optimization using parameters adopting floating or fixed-point real-valued representations--requiring two distinct sets of network parameters. In this paper, we propose a hardware-friendly training method that, contrary to conventional methods, progressively binarizes a singular set of fixed-point network parameters, yielding notable reductions in power and resource utilizations. We use the Intel FPGA SDK for OpenCL development environment to train our progressively binarizing DNNs on an OpenVINO FPGA. We benchmark our training approach on both GPUs and FPGAs using CIFAR-10 and compare it to conventional BNNs.

Corey Lammie, Olga Krestinskaya, Alex James et al.

The quantization of weights to binary states in Deep Neural Networks (DNNs) can replace resource-hungry multiply accumulate operations with simple accumulations. Such Binarized Neural Networks (BNNs) exhibit greatly reduced resource and power requirements. In addition, memristors have been shown as promising synaptic weight elements in DNNs. In this paper, we propose and simulate novel Binarized Memristive Convolutional Neural Network (BMCNN) architectures employing hybrid weight and parameter representations. We train the proposed architectures offline and then map the trained parameters to our binarized memristive devices for inference. To take into account the variations in memristive devices, and to study their effect on the performance, we introduce variations in $R_{ON}$ and $R_{OFF}$. Moreover, we introduce means to mitigate the adverse effect of memristive variations in our proposed networks. Finally, we benchmark our BMCNNs and variation-aware BMCNNs using the MNIST dataset.

Corey Lammie, Wei Xiang, Mostafa Rahimi Azghadi

Recent technological advances have proliferated the available computing power, memory, and speed of modern Central Processing Units (CPUs), Graphics Processing Units (GPUs), and Field Programmable Gate Arrays (FPGAs). Consequently, the performance and complexity of Artificial Neural Networks (ANNs) is burgeoning. While GPU accelerated Deep Neural Networks (DNNs) currently offer state-of-the-art performance, they consume large amounts of power. Training such networks on CPUs is inefficient, as data throughput and parallel computation is limited. FPGAs are considered a suitable candidate for performance critical, low power systems, e.g. the Internet of Things (IOT) edge devices. Using the Xilinx SDAccel or Intel FPGA SDK for OpenCL development environment, networks described using the high-level OpenCL framework can be accelerated on heterogeneous platforms. Moreover, the resource utilization and power consumption of DNNs can be further enhanced by utilizing regularization techniques that binarize network weights. In this paper, we introduce, to the best of our knowledge, the first FPGA-accelerated stochastically binarized DNN implementations, and compare them to implementations accelerated using both GPUs and FPGAs. Our developed networks are trained and benchmarked using the popular MNIST and CIFAR-10 datasets, and achieve near state-of-the-art performance, while offering a >16-fold improvement in power consumption, compared to conventional GPU-accelerated networks. Both our FPGA-accelerated determinsitic and stochastic BNNs reduce inference times on MNIST and CIFAR-10 by >9.89x and >9.91x, respectively.