ACM Open IoT Day - Program

Registration 8:00–9:00

Opening 9:00–9:15

Keynote 1: Feng Zhao (CTO, Haier) 9:15–10:15

IoT: From the Lab to the Real World

The manufacturing industry is going through a digital transformation, from a pure device maker to a hybrid device+service provider. IoT promises to be the enabler for this transformation, connecting at one end the consumers and the other end the manufacturers and their vast resources. I will draw examples from Haier’s practices in smart home and connected factories. Building upon the IoT and other advanced computing technologies, Haier’s U+ Platform has driven the rapid introduction and iteration of new consumer experiences and drastically improved the manufacturing efficiency.

Please see the ACM IoT Day page for more information about the speaker.

Session 1: IoT in the Age of AI 10:30–12:05

Session Chair: Guoliang Xing, The Chinese University of Hong Kong

  • The Deep (Learning) Transformation of Mobile and Embedded Computing 10:30–10:50

    Nic Lane, Oxford University, and Samsung AI Centre

    Mobile and embedded devices increasingly rely on deep neural networks to understand the world -- a formerly impossible feat that would have overwhelmed their system resources just a few years ago. The age of on-device artificial intelligence is upon us; but incredibly, these dramatic changes are just the beginning. Looking ahead, mobile machine learning will extend beyond just classifying categories and perceptual tasks, to roles that alter how every part of the systems stack of smart devices function. This evolutionary step in constrained-resource computing will finally produce devices that meet our expectations in how they can learn, reason and react to the real-world. In this talk, I will briefly discuss the initial breakthroughs that allowed us to reach this point, and outline the next set of open problems we must overcome to bring about this next deep transformation of mobile and embedded computing.

  • Moving to the Edge : Low-latency Face Analysis Engine at Naver Clova 10:55–11:15

    Taewan Kim, NAVER

    Efficient neural network inference on resource-limited edge devices is not a trivial task. In particular, it is common that a couple of neural network models should be run in sequence for applying into real-world applications. In this talk, we will present how to realize such FLOP-intensive neural network models can be run on the resource-limited edge devices with low-latency. In addition, we will introduce our Face Analysis Engine optimized for CPUs.

  • Embracing Ubiquitous Connectivity and AI: The Next Era of IoT 11:20–11:40

    Xue (Steve) Liu, McGill University and Samsung AI Center

    The Internet of Things (IoT) refers to the network of connected devices that communicate seamlessly over the Internet. IoT applications are transforming how we live and work. IoT devices collect/generate a large amount of data. Together with emerging Artificial Intelligence (AI) technologies including cloud AI and on-device AI, these data can be analyzed and contextualized hence enable smart applications to end users. This opens a whole new world of smart applications which allows a richer and closer interaction between people and the environment surrounding them. In this talk, I will illustrate some key potentials and challenges of IoT in this new era.

  • oneM2M Standard-based IoT Platform Mobius and Smart City Data Hub 11:45–12:05

    SeungMyeong Jeong, Korea Electronics Technology Institute (KETI)

    This presentation introduces KETI's activities on IoT and smart city as Mobius and Smart City Data Hub, respectively. Mobius is the oneM2M international standard IoT middleware platform implementation. Mobius is the very first oneM2M certified open source S/W. Including Mobius, KETI is leading the OCEAN open source community providing other IoT platforms and developer tools. Smart City Data Hub is the data-centric platform to solve city problems. Data Hub is being developed within the Smart City Innovation Growth Engine Program in South Korea with two pilot cities.

Lunch 12:00–13:00

Session 2: Intelligible, Fast, Accurate, and Tiny IoT 13:00–14:35

Session Chair: Yunxin Liu, Microsoft Research

  • IoT: digitize the physical world and make sense of it 13:00–13:20

    Sean Ding, Alibaba

    In China, Alibaba Cloud’s IoT now is accelerating the massive growth of IoT in city construction and management by the end-to-end solution, which include Cloud computing platform, including communication management, device management, data analytic platform. Alibaba Cloud’s IOT will also help sensor deployment(NBIOT and Lora), low power network deployment (Lora and NBIOT) in applications like traffic/pipeline network/environment/healthcare monitoring. On top of this platform city manager can build city operation center, helping city builder and regulator to significantly increase the efficiency of city governance with the minimum cost. Meanwhile, Alibaba Cloud’s IoT is also providing the end-to-end solution toward smart home, smart car, smart community, and smart office to change and innovate people’s lifestyle in an intelligent and sustainable way. Alibaba Cloud’s IoT is providing end-to-end solution to smart city. Total solution including voice recognition, nature language processing, music service, and mobile edge computing which running edge computing, as well as including 200 kinds of ICA certified smart sensor. Meanwhile, Alibaba smart city platform provide machine learning platform to build smart application, like smart lighting, smart irrigating, smart parking. With total software and hardware solution, city builder can enable smart city application within days.

  • Enabling Smart IoT 13:25–13:45

    Lucy Cherkasova, ARM Research

    The Internet of Things (IoT) is already comprised of billions interconnected computing devices and this network continues to grow at an amazing pace. The sheer volume of data generated by IoT networks is unprecedented. Harnessing the full potential of IoT demands new approaches along the entire data processing pipeline: how data is created, transferred, processed, ingested, and acted upon. Getting a value from IoT requires making things smarter. Bringing Machine Learning (ML) to the very "edge" of the physical world (where sensing and data collection take place) is a major technological challenge in the coming era of machine intelligence. Arm and its partners continue to develop the machine learning capabilities within microcontroller-based systems. This talk looks at some of the issues and challenges on this path and explores a variety of promising avenues for future innovation (such as TinyML, Federated ML, AutoML, etc.).

  • Wearables, IoT and You 13:50–14:10

    Fahim Kawsar, Nokia Bell Labs and TU Delft

    Wearables and IoT are embracing AI and causing a seismic shift - in that we are observing the emergence of remarkable sensory systems. These transformative sensory systems are capable of understanding us and the world around us, uncovering unprecedented opportunities to help us become a better version of ourselves. In this talk, I will explore the system and algorithmic challenges in modelling and augmenting human behaviour in this new sensory world. I will discuss how mobile, wearable, implantable and IoT devices together with embedded AI can be used as a multi-sensory computational platform to learn, infer, and augment human behaviour and to design ultra-personal computational experience.

  • Designing an Extremely Tiny-Size Deep Face Detector by Recurrent Sharing of Layers 14:15–14:35

    Youngjoon Yoo, NAVER

    In this presentation, we propose a new multi-scale face detector having extremely small number of parameters, less than 0.1 million, as well as achieving comparable performance to deep heavy detectors. While existing multi-scale face detectors extract feature maps with different scale from a single backbone network, our method generates the feature maps by reusing a shared lightweight and shallow backbone network composed of depth-wise convolutions in iterative manner. This iterative sharing of the backbone network significantly reduces the parameter, and also provides the abstract image semantics captured from the higher-stage of the network layers to the first stage feature map. The proposed idea is employed by various model architectures and evaluated by extensive experiments. From the experiments from WIDER FACE dataset, we show that the proposed face detector can handle faces with various scale and conditions, and the performance is comparable to those of much heavier deep face detectors which is few hundreds and tens times heavier in model size and floating point operations.

Session 3: IoT in the Wild 15:00–15:45

Session Chair: Inseok Hwang, IBM Research - Austin

  • Precise indoor positioning solution and IoT technology for large-scale Fleet management 15:00–15:20

    Andrew Jang, POLARIANT

    POLARIANT has developed and commercialized Polarized Light Sensing (PLS), an precise indoor three-dimensional positioning solution using polarized light. The problem that we have solved is that it is difficult to find out the three-dimensional position of the robot and the vehicle that need to be controlled in the indoor space where the GPS does not reach in real time, and it is applied to the mobility market in particular. When the autonomous vehicle travels outdoors, it can be self-propelled by using a various sensors. However, from the moment it enters the large indoor parking lot to the parking slot, they cannot recognized their concrete position in the real time. POLARIANT has recently been acquired by Korean No. 1 mobility platform company, SOCAR which owns over than 13,000 fleets and 5 thousands of parking infrastructures. POLARIANT will accelerate the optimization of its mobility market with SOCAR using precise positioning and embedded system technologies of ours. In the keynote address on the IoT event day, we will introduce the issues related to precise positioning in the mobility market, and introduce the roadmap using IoT technology for large amount of fleet management.

  • From Rhinos to Asteroids: AI/ML on Large Scale Spatiotemporal Data 15:25–15:45

    Raghu Ganti, IBM Watson, NY

    The last decade has seen an explosion in the volume and availability of space-time tagged data from various moving objects. In this talk, I will cover various AI/ML algorithms developed by IBM in the last decade, examples of which include learning animal movement patterns representing poacher attacks, predicting asteroid-asteroid encounters, and detecting human/drug trafficking activities.

Panel Talk 15:45–16:30

Topic: The Convergence of 5G, AI and IoT and its Impact on Human Productivity.

Moderator: Fahim Kawsar, Nokia Bell Labs and TU Delft


Keynote2: Jaeyeon Jung (VP, Samsung Electronics) 16:40–17:40

Democratizing the Internet of Things with SmartThings

SmartThings Cloud connects not only Samsung devices (e.g., home appliances, TVs), but a wide range of connected devices (e.g., sensors, lightbulbs, cameras), providing a single, powerful platform to build IoT experiences to millions of SmartThings users. In this talk, we will discuss how you can be part of this exciting journey—integrating existing products with our ecosystem using SmartThings Schema; creating smart interactions using SmartThings APIs; or connecting your ZigBee or Z-Wave devices with SmartThings hubs. All these can be done through our developer workspace, which can be reached at

Please see the ACM IoT Day page for more information about the speaker.

ACM Open IoT Day Reception 18:00–20:00