Sessions

PhxMobi Emerging Tech Festival

Sponsored by Couchbase

November 14th

How Can You Get Started with Machine Learning?

Ruqiya Bin Safi

Machine Learning Roadmap “From Zero to Hero” With End To End Machine Learning Project.


Battery Technology and the Future of EV Charging

Mark Hanchett

Electric Vehicles shouldn’t just be meeting the status quo, it should be about delivering a radically new driving experience. That’s why ATLIS is creating an advanced charging station and fast charging battery technology all in-house. It’s everything charging should be: Fast, affordable, plug-and-charge and it’s designed to scale. In this discussion, Mark will talk about how ATLIS is able to charge in less than 15 minutes, the future of EV charging stations, and the challenges we face in the Automotive Industry.


AI during the COVID era

Hari Gottipati

The COVID-19 global pandemic changed how we live, work, and shop. It has changed consumer’s behavior and shopping patterns and has caused a widespread disruption the world over. From detection to prevention, remote working to virtual gatherings, and online shopping to streaming content, AI is playing a pivotal role in our day-to-day lives even during the pandemic. This talk covers various AI technologies that are helping us deal with life during and after this pandemic.


Plug and Play AI – How to effortlessly embed cutting edge AI into your business processes and applications

Bindu Reddy

We will review some common deep learning techniques behind automated machine learning and training with less data. We will also then talk about plug and play AI, how it works, and why it is beneficial to apply it to common use-cases and business processes.


Robust and Resilient Safety Control of Autonomous Vehicles

Sze Zheng Yong

Recent research in safety control has leveraged the availability of accurate models to detect impending safety violations and to intervene accordingly. However, there is often a mismatch between the models that are used for algorithm design and the real systems. Moreover, control designs typically assume that the sensor measurements are trustworthy. These modeling discrepancies and the possibility of compromised/spoofed signals, if not proactively considered, will jeopardize safety guarantees, leading to serious damage to safety-critical systems, including autonomous vehicles, and to loss of trust in these technologies.

This talk will provide a brief overview of recent research on robust and resilient safety control for autonomous vehicles. Furthermore, some potential knowledge gaps in this area of safety control will be presented with the goal of increasing our confidence in these technologies.


Deploying Machine Learning Models to Edge Devices

David Brown

Leveraging Machine Learning models ‘at the Edge’ can be a challenge. The possibility of DIL (Disconnected, Intermittent, Low Bandwidth) networks means that your edge device needs to be able to make decisions without assuming it can connect to a server. At the same time, you want to make decisions based on a corpus of data that is generally too large to host on the edge device. Connecting the edge to the cloud and vice versa is an architectural challenge in itself.


Deep Neural Networks (DNN) vs Shallow Neural Networks (SHNN)

Mohammad Farhadi

Object detection using deep neural networks (DNNs) involves a huge amount of computation which impedes its implementation on resource/energy-limited user-end devices. The reason for the success of DNNs is due to having knowledge over all different domains of observed environments. However, we need a limited knowledge of the observed environment at inference time which can be learned using a shallow neural network (SHNN). The “TKD” (Temporal Knowledge Distillation) is a system-level design that proposed to improve the energy consumption of object detection on the user-end device. An SHNN is deployed on the user-end device to detect objects in the observing environment. Also, a knowledge transfer mechanism is implemented to update the SHNN model using the DNN knowledge when there is a change in the object domain. Experiments demonstrate that the energy consumption of the user-end device and the inference time can be improved by 78% and 71% compared with running the deep model on the user-end device.

The proposed technology targets the following commercial applications.

Building Owners / Occupants: Building HVAC accounts for 13% of all energy usage in the United States and nearly 40% of the energy used in buildings. Accurate occupancy detection can reduce energy use in HVAC systems by as much as 30%. Many issues come up when discussing occupancy detection when assessing accurate detection and privacy. This is where the proposed technology solves both, by relying on the vision to provide precise information and on-device analysis to ensure no image privacy data is transmitted.

Municipalities: Traffic light planning is difficult already without accounting for pedestrians. Pedestrians congregating at an intersection can cause safety issues for all vehicles and people. Accurately determining the counts of pedestrians, vehicles, and bicycles is a difficult task for existing technology in the market. The proposed technology allows for simple integration of counting of people, cars, and bikes, and removes concerns that big brother is watching. Additional analysis can be performed by providing count data – if pedestrian counts are not changing, there may be a need to modify traffic patterns or alert first responders. Starting with simple, actionable data is the first step in this process.


EdgeX Foundry – Using an open source IoT platform to build edge solutions

Jim White

EdgeX Foundry is a vendor-neutral, open-source, hardware, and OS agnostic Linux Foundation project to create a common open platform for IoT edge computing systems. EdgeX is a software framework that enables you to rapidly connect the “things” in your IoT environment to your enterprise or cloud systems. It has the support of over 70 companies, has been adopted by the likes of Accenture, ThunderSoft, HP, Tibco, and others, and has had more than 7 million container downloads.

In this session, learn how EdgeX Foundry can help in the development of interoperable and distributable edge/fog/IoT solutions. See how to get and deploy EdgeX. Learn how to connect a multitude of sensors and devices and get data from the edge to your cloud or enterprise platforms. See how EdgeX provides the means to provide local edge analytics and explore how you can customize/extend EdgeX to suit your IoT use cases.


5G Radio Access Networks Move to the Cloud

Kalyan Subramanian

As 5G RANs roll out with the promise of increased capacity and speed, dis-aggregation of compute elements to cloud-based implementation is the inevitable solution. This talk will provide a brief overview of emerging areas in 5G and opening up the ecosystem to exciting new innovation possibilities in areas like AI & ML.


Please stop building killer robots!

Bryce Howitson

AI & ML are the future of all technology. But it’s not new and that history (plus popular culture) has taught us to be skeptical of how AI will negatively impact humanity. Killer robots are a pretty popular theme in science fiction after all.

I’ll share my experience living as an imperfect human and how AI is making me better. I’ll also cover case studies of AI in healthcare, job placement, and elsewhere that augment task people do. After this talk, you will start seeing AI applications that already exist and learn to identify opportunities to create AI/ML solutions to improve humans’ relationship with technology.


Who’s to blame? Inferring driver belief dynamics through large-scale traffic data

Dr. Yi Ren and Dr. Wenlong Zhang

Vehicle interactions are incomplete-information differential games where the belief and physical states of agents are entangled and evolve over time. For a given observation of a traffic scene, e.g., an accident, can we trace back the belief states of agents, i.e., what everyone was thinking all the way up to the accident? Can we identify prosocial vs. competitive driving behaviors through surveillance? In this talk, we answer these questions through a discussion that connects behavioral economics, game theory, optimal control theory, and machine learning.


Why building Responsible AI matters!

Usha Jagannathan

We will discuss why building responsible AI is a business imperative. The discussion will focus on why and how organizations apply ethical principles into actionable practice when designing, deploying, and evaluating AI systems to make informed business decisions.


A Technology Roadmap for Contactless Commerce

Katina Michael

COVID-19 has heightened awareness about the requirement for contactless transactions. But the question is: were we always set on this course to begin with? Has the pandemic transmission merely acted to speed up the inevitable? Who wants to carry cash in their wallets and purses, or even cards or mobile phones? This presentation will take a quick snapshot through the history of the future possibilities. What are some of the future scenarios consumers might be asked to consider to futureproof? This presentation will rely on two decades of research in the luggable, wearable, and implantable spaces, providing evidence for future contexts.