Due to the limited computation and storage resources of Internet of Things (IoT) devices, many emerging intelligent applications based on deep learning techniques heavily depend on cloud computing for computation and storage. However, cloud computing faces technical issues with long latency, poor reliability, and weak privacy, resulting in the need for on-device computation and storage. Also, on-device computation is essential for many time-critical applications, which require real-time data processing and energy-efficient. Furthermore, the escalating requirements for on-device processing are driven by network bandwidth limitations and consumer anticipations concerning data privacy and user experience. In the realm of computing, there is a growing interest in exploring novel technologies that can facilitate ongoing advancements in performance. Of the various prospective avenues, the field of neuromorphic computing has garnered significant recognition as a crucial means to achieve fast and energy-efficient machine intelligence applications for IoT devices. The programming of neuromorphic computing hardware typically involves the construction of a spiking neural network (SNN) capable of being deployed onto the designated neuromorphic hardware. This dissertation presents a range of methodologies aimed at enhancing the precision and energy efficiency of SNNs. To be more precise, these advancements are achieved by incorporating four essential methods. The first method is the quantization of neural networks through knowledge distillation. This work introduces a quantization technique that effectively reduces the computational and storage resource requirements of a model while minimizing the loss of accuracy. To further enhance the reduction of quantization errors, the second method introduces a novel quantization-aware training algorithm specifically designed for training quantized spiking neural network (SNN) models intended for execution on the Loihi chip, a specialized neuromorphic computing chip. SNNs generally exhibit lower accuracy performance compared to deep neural networks (DNNs). The third approach introduces a DNN-SNN co-learning algorithm, which enhances the performance of SNN models by leveraging knowledge obtained from DNN models. The design of the neural architecture plays a vital role in enhancing the accuracy and energy efficiency of an SNN model. The fourth method presents a novel neural architecture search algorithm specifically tailored for SNNs on the Loihi chip. The method selects an optimal architecture based on gradients induced by the architecture at initialization across different data samples without the need for training the architecture. To demonstrate the effectiveness and performance across diverse machine intelligence applications, our methods are evaluated through (i) image classification, (ii) spectrum sensing, and (iii) modulation symbol detection. / Doctor of Philosophy / In the emerging Internet of Things (IoT), our everyday devices, from smart home gadgets to wearables, can autonomously make intelligent decisions. However, due to their limited computing power and storage, many IoT devices heavily depend on cloud computing, which brings along issues like slow response times, privacy concerns, and unreliable connections. Neuromorphic computing is a recognized and crucial approach for achieving fast and energy-efficient machine intelligence applications in IoT devices. Inspired by the human brain's neural networks, this cutting-edge approach allows devices to perform complex tasks efficiently and in real-time. The programming of this neuromorphic hardware involves creating spiking neural networks (SNNs). This dissertation presents several innovative methods to improve the precision and energy efficiency of these SNNs. Firstly, a technique called "quantization" reduces the computational and storage requirements of models without sacrificing accuracy. Secondly, a unique training algorithm is designed to enhance the performance of SNN models. Thirdly, a clever co-learning algorithm allows SNN models to learn from traditional deep neural networks (DNNs), further improving their accuracy. Lastly, a novel neural architecture search algorithm finds the best architecture for SNNs on the designated neuromorphic chip, without the need for extensive training. By making IoT devices smarter and more efficient, neuromorphic computing brings us closer to a world where our gadgets can perform intelligent tasks independently, enhancing convenience and privacy for users across the globe.

Quantization of Neural Networks is popular technique for adopting computation intensive Deep Learning applications to edge devices. In this work, low bit mixed precision quantization of FPN-Resnet18 model trained for the task of semantic segmentation is explored using Cityscapes and Arriver datasets. The Hessian information of each layer in the model is used to determine the bit precision for each layer and in some experiments the bit precision for the layers are determined randomly. The networks are quantization-aware trained with bit combinations 2, 4 and 8. The results obtained for both Cityscapes and Arriver datasets show that the quantization-aware trained networks with the low bit mixed precision technique offer a performance at par with the 8-bit quantization-aware trained networks and the segmentation performance degrades when the network activations are quantized below 8 bits. Also, it was found that the usage of the Hessian information had little effect on the network’s performance. / Kvantisering av Neurala nätverk är populär teknik för att införa beräknings-intensiva Deep Learning -applikationer till edge-enheter. I detta arbete utforskas låg bitmixad precisionskvantisering av FPN-Resnet18-modellen som är utbildad för uppgiften för semantisk segmentering med hjälp av Cityscapes och Arriverdatauppsättningar. Hessisk information från varje lager i modellen, används för att bestämma bitprecisionen för respektive lager. I vissa experiment bestäms bitprecision för skikten slumpmässigt. Nätverken är kvantiserings medvetna utbildade med bitkombinationer 2, 4 och 8. Resultaten som erhållits för både Cityscapes och Arriver datauppsättningar visar att de kvantiserings medvetna utbildade nätverken med lågbit blandad precisionsteknik erbjuder en prestanda i nivå med 8-bitars kvantiseringsmedvetna utbildade nätverk och segmenteringens prestationsgrader när nätverksaktiveringarna kvantiseras under 8 bitar. Det visade sig också att användningen av hessisk information hade liten effekt på nätets prestanda.

Emerging machine learning (ML) techniques are showing great potential in prediction performance. However, research and development is often conducted in an environment with extensive computational resources and blinded by prediction performance. In reality, computational resources might be contained on constrained hardware where energy and memory consumption must be restrained. Furthermore, shortages of sufficiently large datasets for ML is a frequent problem, combined with the cost of data retention. This generates a significant demand for sustainable ML. With sustainable ML, practitioners can train ML models on less data, which reduces memory and energy consumption during the training process. To explore solutions to these problems, this thesis dives into several techniques that have been introduced in the literature to achieve energy-savings when training machine learning models. These techniques include Quantization-Aware Training, Model Distillation, Quantized Distillation, Continual Learning and a deeper dive into Siamese Neural Networks (SNNs), one of the most promising techniques for sustainability. Empirical evaluations are conducted using several datasets to illustrate the potential of these techniques and their contribution to sustainable ML. The findings of this thesis show that the energy-saving techniques could be leveraged in some cases to make machine learning models more manageable and sustainable whilst not compromising significant model prediction performance. In addition, the deeper dive into SNNs shows that SNNs can outperform standard classification networks, under both the standard multi-class classification case and the Continual Learning case, whilst being trained on significantly less data. / Maskininlärning har i den senaste tidens forskning visat stor potential och hög precision inom klassificering. Forskning, som ofta bedrivs i en miljö med omfattande beräkningsresurser, kan lätt bli förblindad av precision. I verkligheten är ofta beräkningsresurser lokaliserade på hårdvara där energi- och minneskapacitet är begränsad. Ytterligare ett vanligt problem är att uppnå en tillräckligt stor datamängd för att uppnå önskvärd precision vid träning av maskininlärningsmodeller. Dessa problem skapar en betydande efterfrågan av hållbar maskininlärning. Hållbar maskininlärning har kapaciteten att träna modeller på en mindre datamängd, vilket minskar minne- och energiförbrukning under träningsprocessen. För att utforska hållbar maskininlärning analyserar denna avhandling Quantization-Aware Training, Model Distillation, Quantized Distillation, Continual Learning och en djupare evaluering av Siamesiska Neurala Nätverk (SNN), en av de mest lovande teknikerna inom hållbar maskininlärning. Empiriska utvärderingar utfördes med hjälp av flera olika datamängder för att illustrera potentialen hos dessa tekniker. Resultaten visar att energibesparingsteknikerna kan utnyttjas för att göra maskininlärningsmodeller mer hållbara utan att kompromissa för precision. Dessutom visar undersökningen av SNNs att de kan överträffa vanliga neurala nätverk, med och utan Continual Learning, även om de tränas på betydligt mindre data.