“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

The appearance of cars has raised materialistic civilization and living standard to an unprecedented level. Today, it is hard to imagine how we human beings can live without cars. Yet, motor vehicles can cause a great number of deaths and injuries as well as considerable economic losses, which have constituted the global burden. Understanding of the occurrence and development of road traffic injuries will contribute to the prevention and control of crash and to the implementation of "everybody has the right to enjoy health" proposed by WHO.

Road traffic injuries.

Energy disaggregation of appliances using non-intrusive load monitoring (NILM) represents a set of signal and information processing methods used for appliance-level information extraction out of a meter’s total or aggregate load. Large-scale deployments of smart meters worldwide and the availability of large amounts of data, motivates the shift from traditional source separation and Hidden Markov Model-based NILM towards data-driven NILM methods. Furthermore, we address the potential for scalable NILM roll-out by tackling disaggregation complexity as well as disaggregation on houses which have not been ’seen’ before by the network, e.g., during training. In this paper, we focus on low rate NILM (with active power meter measurements sampled between 1-60 seconds) and present two different neural network architectures, one, based on convolutional neural network, and another based on gated recurrent unit, both of which classify the state and estimate the average power consumption of targeted appliances. Our proposed designs are driven by the need to have a well-trained generalised network which would be able to produce accurate results on a house that is not present in the training set, i.e., transferability. Performance results of the designed networks show excellent generalization ability and improvement compared to the state of the art.

Transferability of Neural Network Approaches for Low-rate Energy Disaggregation

Recent advances in machine learning (ML) have produced KiloByte-size models that can directly run on constrained IoT devices. This approach avoids expensive communication between IoT devices and the cloud, thereby enabling energy-efficient real-time analytics. However, ML models are expressed typically in floating-point, and IoT hardware typically does not support floating-point. Therefore, running these models on IoT devices requires simulating IEEE-754 floating-point using software, which is very inefficient. We present SeeDot, a domain-specific language to express ML inference algorithms and a compiler that compiles SeeDot programs to fixed-point code that can efficiently run on constrained IoT devices. We propose 1) a novel compilation strategy that reduces the search space for some key parameters used in the fixed-point code, and 2) new efficient implementations of expensive operations. SeeDot compiles state-of-the-art KB-sized models to various microcontrollers and low-end FPGAs. We show that SeeDot outperforms 1) software emulation of floating-point (Arduino), 2) high-bitwidth fixed-point (MATLAB), 3) post-training quantization (TensorFlow-Lite), and 4) floating- and fixed-point FPGA implementations generated using high-level synthesis tools.

Compiling KB-sized machine learning models to tiny IoT devices

Anomaly detection has gained considerable attention in the past couple of years. Emerging technologies, such as the Internet of Things (IoT), are known to be among the most critical sources of data streams that produce massive amounts of data continuously from numerous applications. Examining these collected data to detect suspicious events can reduce functional threats and avoid unseen issues that cause downtime in the applications. Due to the dynamic nature of the data stream characteristics, many unresolved problems persist. In the existing literature, methods have been designed and developed to evaluate certain anomalous behaviors in IoT data stream sources. However, there is a lack of comprehensive studies that discuss all the aspects of IoT data processing. Thus, this paper attempts to fill this gap by providing a complete image of various state-of-the-art techniques on the major problems and core challenges in IoT data. The nature of data, anomaly types, learning mode, window model, datasets, and evaluation criteria are also presented. Research challenges related to data evolving, feature-evolving, windowing, ensemble approaches, nature of input data, data complexity and noise, parameters selection, data visualizations, heterogeneity of data, accuracy, and large-scale and high-dimensional data are investigated. Finally, the challenges that require substantial research efforts and future directions are summarized.

A Review of Machine Learning and Deep Learning Techniques for Anomaly Detection in IoT Data

The proliferation of Internet of Things (IoT) devices has led to the deployment of various types of sensors in the homes, offices, buildings, lawns, cities, and even in agricultural farms. Since IoT applications rely on the fidelity of data reported by the sensors, it is important to detect a faulty sensor and isolate the cause of the fault. Existing fault detection techniques demand sensor domain knowledge along with the contextual information and historical data from similar near-by sensors. However, detecting a sensor fault by analyzing just the sensor data is non-trivial since a faulty sensor reading could mimic non-faulty sensor data. This paper presents a novel primitive, which we call the Fall-curve - a sensor's voltage response when the power is turned off - that can be used to characterize sensor faults. The Fall-curve constitutes a unique signature independent of the phenomenon being monitored which can be used to identify the sensor and determine whether the sensor is correctly operating. We have empirically evaluated the Fall-curve technique on a wide variety of analog and digital sensors. We have also been running this system live in a few agricultural farms, with over 20 IoT devices. We were able to detect and isolate faults with an accuracy over 99%, which would have otherwise been hard to detect only by observing measured sensor data.

Fall-curve: A novel primitive for IoT Fault Detection and Isolation

Driver inattention is one of the leading causes of vehicle crashes and incidents worldwide. Driver inattention includes driver fatigue leading to drowsiness and driver distraction, say due to use of cellphone or rubbernecking, all of which leads to a lack of situational awareness. Hitherto, techniques presented to monitor driver attention evaluated factors such as fatigue and distraction independently. However, in order to develop a robust driver attention monitoring system all the factors affecting driver’s attention needs to be analyzed holistically. In this paper, we present AutoRate, a system that leverages front camera of a windshield-mounted smartphone to monitor driver’s attention by combining several features. We derive a driver attention rating by fusing spatio-temporal features based on the driver state and behavior such as head pose, eye gaze, eye closure, yawns, use of cellphones, etc.We perform extensive evaluation of AutoRateon real-world driving data and also data from controlled, static vehicle settings with 30 drivers in a large city. We compare AutoRate’s automatically-generated rating with the scores given by 5 human annotators. Further, we compute the agreement between AutoRate’s rating and human annotator rating using kappa coefficient. AutoRate’s automatically-generated rating has an overall agreement of 0.87 with the ratings provided by 5 human annotators on the static dataset.

AutoRate: How attentive is the driver?

Road safety is a major public health issue the world over. Many studies have found that the primary factors responsible for road accidents center on the driver and her/his driving. Hence, there is the need to monitor driver's state and her/his driving, with a view to providing effective feedback. Our proposed demo is of HAMS, a windshield-mounted, smartphone-based system that uses the front camera to monitor the driver and back camera to monitor her/his driving behaviour. The objective of HAMS is to provide ADAS-like functionality with low-cost devices that can be retrofitted onto the large installed base of vehicles that lack specialized and expensive sensors such as LIDAR and RADAR. Our demo would show HAMS in action on an Android smartphone to monitor the state of the driver, specifically such as drowsiness, distraction and gaze, and vehicle ranging, lane detection running on pre-recorded videos from drives.

HAMS: Driver and Driving Monitoring using a Smartphone

Road safety is a major public health issue across the globe and over two-thirds of the road accidents occur at nighttime under low-light conditions or darkness. The state of the driver and her/his actions are the key factors impacting road safety. How can we monitor these in a cost-effective manner and in low-light conditions? RGB cameras present in smartphones perform poorly in low-lighting conditions due to lack of information captured. Hence, existing monitoring solutions rely upon specialized hardware such as infrared cameras or thermal cameras in low-light conditions, but are limited to only high-end vehicles owing to the cost of the hardware. We present InSight, a windshield-mounted smartphone-based system that can be retrofitted to the vehicle to monitor the state of the driver, specifically driver fatigue (based on frequent yawning and eye closure) and driver distraction (based on their direction of gaze). Challenges arise from designing an accurate, yet low-cost and non-intrusive system to continuously monitor the state of the driver. In this paper, we present two novel and practical approaches for continuous driver monitoring in low-light conditions: (i) Image synthesis: enabling monitoring in low-light conditions using just the smartphone RGB camera by synthesizing a thermal image from RGB with a Generative Adversarial Network, and (ii) Near-IR LED: using a low-cost near-IR (NIR) LED attachment to the smartphone, where the NIR LED acts as a light source to illuminate the driver's face, which is not visible to the human eyes, but can be captured by standard smartphone cameras without any specialized hardware. We show that the proposed techniques can capture the driver's face accurately in low-lighting conditions to monitor driver's state. Further, since NIR and thermal imagery is significantly different than RGB images, we present a systematic approach to generate labelled data, which is used to train existing computer vision models. We present an extensive evaluation of both the approaches with data collected from 15 drivers in controlled basement area and on real roads in low-light conditions. The proposed NIR LED setup has an accuracy (Fl-score) of 85% and 93.8% in detecting driver fatigue and distraction, respectively in low-light.

InSight: Monitoring the State of the Driver in Low-Light Using Smartphones

Sensor data quality plays a fundamental role in increasing the adoption of IoT devices for environmental data collection. Due to the nature of the deployment, i.e., in-the-wild and in harsh environments, coupled with limitations of low-cost components, sensors are prone to failures. A significant fraction of faults result from drift and catastrophic faults in sensors' sensing components leading to serious data inaccuracies. However, it is challenging to detect faults by analyzing just the sensor data as a faulty sensor data can mimic non-faulty data and an anomalous sensor reading need not represent a faulty data. Existing data-centric approaches rely on additional contextual information or sensor redundancy to detect such faults. This paper presents a systematic approach to detect faults and drifts, by devising a novel sensor fingerprint called CurrentSense. CurrentSense captures the electrical characteristics of the hardware components in a sensor, with working, drifted, and faulty sensors having distinct fingerprints. This fingerprint is used to determine the sensors' health, and compensate for drift or diagnose catastrophic faults without any contextual information. The CurrentSense approach is non-intrusive, and can be applied to a wide variety of environmental sensors. We show the working of the proposed approach with the help of air pollution sensors. We perform an extensive evaluation in both controlled setup and real-world deployments with 51 sensors across multiple cities for 8 months period. Our approach outperforms existing anomaly detectors and can detect and isolate faults with an F1 score of 98% and compensate for sensor drift errors by 86%.

https://dl.acm.org/doi/pdf/10.1145/3450268.3453535

Akshay Uttama Nambi

Papers

Fall-curve: A novel primitive for IoT Fault Detection and Isolation

AutoRate: How attentive is the driver?

HAMS: Driver and Driving Monitoring using a Smartphone

InSight: Monitoring the State of the Driver in Low-Light Using Smartphones

CurrentSense: A novel approach for fault and drift detection in environmental IoT sensors