Akhil Mathur
YOU?
Author Swipe
View article: CroSSL: Cross-modal Self-Supervised Learning for Time-series through Latent Masking
CroSSL: Cross-modal Self-Supervised Learning for Time-series through Latent Masking Open
Limited availability of labeled data for machine learning on multimodal time-series extensively hampers progress in the field. Self-supervised learning (SSL) is a promising approach to learn data representations without relying on labels. …
View article: Balancing Continual Learning and Fine-tuning for Human Activity Recognition
Balancing Continual Learning and Fine-tuning for Human Activity Recognition Open
Wearable-based Human Activity Recognition (HAR) is a key task in human-centric machine learning due to its fundamental understanding of human behaviours. Due to the dynamic nature of human behaviours, continual learning promises HAR system…
View article: SensiX++: Bringing MLOps and Multi-tenant Model Serving to Sensory Edge Devices
SensiX++: Bringing MLOps and Multi-tenant Model Serving to Sensory Edge Devices Open
We present SensiX++, a multi-tenant runtime for adaptive model execution with integrated MLOps on edge devices, e.g., a camera, a microphone, or IoT sensors. SensiX++ operates on two fundamental principles: highly modular componentisation …
View article: CroSSL: Cross-modal Self-Supervised Learning for Time-series through Latent Masking
CroSSL: Cross-modal Self-Supervised Learning for Time-series through Latent Masking Open
Limited availability of labeled data for machine learning on multimodal time-series extensively hampers progress in the field. Self-supervised learning (SSL) is a promising approach to learning data representations without relying on label…
View article: Tiny, Always-on, and Fragile: Bias Propagation through Design Choices in On-device Machine Learning Workflows
Tiny, Always-on, and Fragile: Bias Propagation through Design Choices in On-device Machine Learning Workflows Open
Billions of distributed, heterogeneous, and resource constrained IoT devices deploy on-device machine learning (ML) for private, fast, and offline inference on personal data. On-device ML is highly context dependent and sensitive to user, …
View article: Kaizen: Practical Self-supervised Continual Learning with Continual Fine-tuning
Kaizen: Practical Self-supervised Continual Learning with Continual Fine-tuning Open
Self-supervised learning (SSL) has shown remarkable performance in computer vision tasks when trained offline. However, in a Continual Learning (CL) scenario where new data is introduced progressively, models still suffer from catastrophic…
View article: Enhancing Efficiency in Multidevice Federated Learning through Data Selection
Enhancing Efficiency in Multidevice Federated Learning through Data Selection Open
Ubiquitous wearable and mobile devices provide access to a diverse set of data. However, the mobility demand for our devices naturally imposes constraints on their computational and communication capabilities. A solution is to locally lear…
View article: A Taxonomy of Noise in Voice Self-reports while Running
A Taxonomy of Noise in Voice Self-reports while Running Open
Smart earables offer great opportunities for conducting ubiquitous computing research. This paper shares its reflection on collecting self-reports from runners using the microphone on the smart eSense earbud device. Despite the advantages …
View article: FLAME
FLAME Open
Federated Learning (FL) enables distributed training of machine learning models while keeping personal data on user devices private. While we witness increasing applications of FL in the area of mobile sensing, such as human activity recog…
View article: Orchestra: Unsupervised Federated Learning via Globally Consistent Clustering
Orchestra: Unsupervised Federated Learning via Globally Consistent Clustering Open
Federated learning is generally used in tasks where labels are readily available (e.g., next word prediction). Relaxing this constraint requires design of unsupervised learning techniques that can support desirable properties for federated…
View article: FLOWER: A FRIENDLY FEDERATED LEARNING FRAMEWORK
FLOWER: A FRIENDLY FEDERATED LEARNING FRAMEWORK Open
Open-Source, mobile-friendly Federated Learning framework
View article: FLAME: Federated Learning Across Multi-device Environments
FLAME: Federated Learning Across Multi-device Environments Open
Federated Learning (FL) enables distributed training of machine learning models while keeping personal data on user devices private. While we witness increasing applications of FL in the area of mobile sensing, such as human activity recog…
View article: ColloSSL: Collaborative Self-Supervised Learning for Human Activity Recognition
ColloSSL: Collaborative Self-Supervised Learning for Human Activity Recognition Open
A major bottleneck in training robust Human-Activity Recognition models (HAR) is the need for large-scale labeled sensor datasets. Because labeling large amounts of sensor data is an expensive task, unsupervised and semi-supervised learnin…
View article: Tiny, always-on and fragile: Bias propagation through design choices in on-device machine learning workflows
Tiny, always-on and fragile: Bias propagation through design choices in on-device machine learning workflows Open
Billions of distributed, heterogeneous and resource constrained IoT devices deploy on-device machine learning (ML) for private, fast and offline inference on personal data. On-device ML is highly context dependent, and sensitive to user, u…
View article: FRuDA: Framework for Distributed Adversarial Domain Adaptation
FRuDA: Framework for Distributed Adversarial Domain Adaptation Open
Breakthroughs in unsupervised domain adaptation (uDA) can help in adapting models from a label-rich source domain to unlabeled target domains. Despite these advancements, there is a lack of research on how uDA algorithms, particularly thos…
View article: Device or User
Device or User Open
We are witnessing a trend of users owning multiple data-generating wearable and IoT devices that continuously capture sensor data pertaining to a user's activities and context. Federated Learning is a potential technique to derive meaningf…
View article: Characterising the Role of Pre-Processing Parameters in Audio-based Embedded Machine Learning
Characterising the Role of Pre-Processing Parameters in Audio-based Embedded Machine Learning Open
When deploying machine learning (ML) models on embedded and IoT devices, performance encompasses more than an accuracy metric: inference latency, energy consumption, and model fairness are necessary to ensure reliable performance under het…
View article: Chronic Pain Protective Behavior Detection with Deep Learning
Chronic Pain Protective Behavior Detection with Deep Learning Open
In chronic pain rehabilitation, physiotherapists adapt physical activity to patients’ performance based on their expression of protective behavior, gradually exposing them to feared but harmless and essential everyday activities. As rehabi…
View article: Low-Power Audio Keyword Spotting Using Tsetlin Machines
Low-Power Audio Keyword Spotting Using Tsetlin Machines Open
The emergence of artificial intelligence (AI) driven keyword spotting (KWS) technologies has revolutionized human to machine interaction. Yet, the challenge of end-to-end energy efficiency, memory footprint and system complexity of current…
View article: On-device Federated Learning with Flower
On-device Federated Learning with Flower Open
Federated Learning (FL) allows edge devices to collaboratively learn a shared prediction model while keeping their training data on the device, thereby decoupling the ability to do machine learning from the need to store data in the cloud.…
View article: A first look into the carbon footprint of federated learning
A first look into the carbon footprint of federated learning Open
Despite impressive results, deep learning-based technologies also raise severe privacy and environmental concerns induced by the training procedure often conducted in data centers. In response, alternatives to centralized training such as …
View article: Low-Power Audio Keyword Spotting using Tsetlin Machines
Low-Power Audio Keyword Spotting using Tsetlin Machines Open
The emergence of Artificial Intelligence (AI) driven Keyword Spotting (KWS) technologies has revolutionized human to machine interaction. Yet, the challenge of end-to-end energy efficiency, memory footprint and system complexity of current…
View article: SensiX: A Platform for Collaborative Machine Learning on the Edge
SensiX: A Platform for Collaborative Machine Learning on the Edge Open
The emergence of multiple sensory devices on or near a human body is uncovering new dynamics of extreme edge computing. In this, a powerful and resource-rich edge device such as a smartphone or a Wi-Fi gateway is transformed into a persona…
View article: Leveraging Activity Recognition to Enable Protective Behavior Detection in Continuous Data
Leveraging Activity Recognition to Enable Protective Behavior Detection in Continuous Data Open
Protective behavior exhibited by people with chronic pain (CP) during physical activities is the key to understanding their physical and emotional states. Existing automatic protective behavior detection (PBD) methods rely on pre-segmentat…
View article: Can Federated Learning Save The Planet?
Can Federated Learning Save The Planet? Open
Despite impressive results, deep learning-based technologies also raise severe privacy and environmental concerns induced by the training procedure often conducted in data centers. In response, alternatives to centralized training such as …
View article: Flower: A Friendly Federated Learning Research Framework
Flower: A Friendly Federated Learning Research Framework Open
Federated Learning (FL) has emerged as a promising technique for edge devices to collaboratively learn a shared prediction model, while keeping their training data on the device, thereby decoupling the ability to do machine learning from t…
View article: Libri-Adapt: a New Speech Dataset for Unsupervised Domain Adaptation
Libri-Adapt: a New Speech Dataset for Unsupervised Domain Adaptation Open
This paper introduces a new dataset, Libri-Adapt, to support unsupervised\ndomain adaptation research on speech recognition models. Built on top of the\nLibriSpeech corpus, Libri-Adapt contains English speech recorded on mobile and\nembedd…
View article: Resource Characterisation of Personal-Scale Sensing Models on Edge Accelerators
Resource Characterisation of Personal-Scale Sensing Models on Edge Accelerators Open
1
View article: EDUQA: Educational Domain Question Answering System Using Conceptual Network Mapping
EDUQA: Educational Domain Question Answering System Using Conceptual Network Mapping Open
Most of the existing question answering models can be largely compiled into\ntwo categories: i) open domain question answering models that answer generic\nquestions and use large-scale knowledge base along with the targeted web-corpus\nret…
View article: Automatic Detection of Protective Behavior in Chronic Pain Physical Rehabilitation: A Recurrent Neural Network Approach.
Automatic Detection of Protective Behavior in Chronic Pain Physical Rehabilitation: A Recurrent Neural Network Approach. Open
In chronic pain physical rehabilitation, physiotherapists adapt physical activity to patients' performance especially based on the expression of protective behavior, gradually exposing them to feared but harmless and essential everyday act…