SMMF: Square-Matricized Momentum Factorization for Memory-Efficient Optimization

We propose SMMF (Square-Matricized Momentum Factorization), a memory-efficient optimizer that reduces the memory requirement of the widely used adaptive learning rate optimizers, such as Adam, by up to 96%. SMMF enables flexible and efficient factorization of an arbitrary rank (shape) of the first and second momentum tensors during optimization, based on the proposed square-matricization and one-time single matrix factorization. From this, it becomes effectively applicable to any rank (shape) of momentum tensors, i.e., bias, matrix, and any rank-d tensors, prevalent in various deep model architectures, such as CNNs (high rank) and Transformers (low rank), in contrast to existing memory-efficient optimizers that applies only to a particular (rank-2) momentum tensor, e.g., linear layers. We conduct a regret bound analysis of SMMF, which shows that it converges similarly to non-memory-efficient adaptive learning rate optimizers, such as AdamNC, providing a theoretical basis for its competitive optimization capability. In our experiment, SMMF takes up to 96% less memory compared to state-of-the-art memory efficient optimizers, e.g., Adafactor, CAME, and SM3, while achieving comparable model performance on various CNN and Transformer tasks.

Designing Extremely Memory-Efficient CNNs for On-device Vision Tasks

In this paper, we introduce a memory-efficient CNN (convolutional neural network), which enables resource-constrained low-end embedded and IoT devices to perform on-device vision tasks, such as image classification and object detection, using extremely low memory, i.e., only 63 KB on ImageNet classification. Based on the bottleneck block of MobileNet, we propose three design principles that significantly curtail the peak memory usage of a CNN so that it can fit the limited KB memory of the low-end device. First, 'input segmentation' divides an input image into a set of patches, including the central patch overlapped with the others, reducing the size (and memory requirement) of a large input image. Second, 'patch tunneling' builds independent tunnel-like paths consisting of multiple bottleneck blocks per patch, penetrating through the entire model from an input patch to the last layer of the network, maintaining lightweight memory usage throughout the whole network. Lastly, 'bottleneck reordering' rearranges the execution order of convolution operations inside the bottleneck block such that the memory usage remains constant regardless of the size of the convolution output channels. The experiment result shows that the proposed network classifies ImageNet with extremely low memory (i.e., 63 KB) while achieving competitive top-1 accuracy (i.e., 61.58\%). To the best of our knowledge, the memory usage of the proposed network is far smaller than state-of-the-art memory-efficient networks, i.e., up to 89x and 3.1x smaller than MobileNet (i.e., 5.6 MB) and MCUNet (i.e., 196 KB), respectively.

CAFO: Feature-Centric Explanation on Time Series Classification

In multivariate time series (MTS) classification, finding the important features (e.g., sensors) for model performance is crucial yet challenging due to the complex, high-dimensional nature of MTS data, intricate temporal dynamics, and the necessity for domain-specific interpretations. Current explanation methods for MTS mostly focus on time-centric explanations, apt for pinpointing important time periods but less effective in identifying key features. This limitation underscores the pressing need for a feature-centric approach, a vital yet often overlooked perspective that complements time-centric analysis. To bridge this gap, our study introduces a novel feature-centric explanation and evaluation framework for MTS, named CAFO (Channel Attention and Feature Orthgonalization). CAFO employs a convolution-based approach with channel attention mechanisms, incorporating a depth-wise separable channel attention module (DepCA) and a QR decomposition-based loss for promoting feature-wise orthogonality. We demonstrate that this orthogonalization enhances the separability of attention distributions, thereby refining and stabilizing the ranking of feature importance. This improvement in feature-wise ranking enhances our understanding of feature explainability in MTS. Furthermore, we develop metrics to evaluate global and class-specific feature importance. Our framework's efficacy is validated through extensive empirical analyses on two major public benchmarks and real-world datasets, both synthetic and self-collected, specifically designed to highlight class-wise discriminative features. The results confirm CAFO's robustness and informative capacity in assessing feature importance in MTS classification tasks. This study not only advances the understanding of feature-centric explanations in MTS but also sets a foundation for future explorations in feature-centric explanations.

Intermittent Learning: On-Device Machine Learning on Intermittently Powered System

In this paper, we introduce the concept of intermittent learning, which enables energy harvested computing platforms to execute certain classes of machine learning tasks. We identify unique challenges to intermittent learning relating to the data and application semantics of machine learning tasks. To address these challenges, we devise an algorithm that determines a sequence of actions to achieve the desired learning objective under tight energy constraints. We further increase the energy efficiency of the system by proposing three heuristics that help an intermittent learner decide whether to learn or discard training examples at run-time. In order to provide a probabilistic bound on the completion of a learning task, we perform an energy event-based analysis that helps us analyze intermittent learning systems where the uncertainty lies in both energy and training example generation. We implement and evaluate three intermittent learning applications that learn the air quality, human presence, and vibration using solar, RF, and kinetic energy harvesters, respectively. We demonstrate that the proposed framework improves the energy efficiency of a learner by up to 100% and cuts down the number of learning examples by up to 50% when compared to state-of-the-art intermittent computing systems without our framework.