QIF 2019 North America

2019 North America Program Details

We believe that research and development is the key to harnessing the power of imagination and to discovering new possibilities. We are excited to announce a new kind of Fellowship that promotes Qualcomm’s core values of innovation, execution and partnership. Our goal is to enable students to pursue their futuristic innovative ideas.

Finalist Selections
Finalist Instructions
Abstract Selections
Participating Universities
Areas of Interest
Timeline

The following 37 proposals have been selected as Finalists. Congratulations!

Please see the Finalists Instructions tab for details on next steps

Qualcomm Research is a division of Qualcomm Technologies, Inc.

CalTech

A nearly optimal self-tuning rate-constrained controller

CMU

Do-­it-­Yourself-­Locally: An IoT architecture For Localized Data Control for Privacy and Security

CMU

Robust and Scalable Machine Learning on the Edge

CMU

MADS: A Machine Learning Assisted Diagnosis System

Cornell

Exploiting Spatial Structure for Dynamic DNN Compression

Cornell

PPAC: In-Memory Accelerator for Matrix-Vector Products

Georgia Tech

Enabling Low-Cost Hardware Security against Side-Channels

Georgia Tech

Hardware-Software Co-design for Deep Learning Acceleration

Georgia Tech

Enabling Efficient Training of Deep Neural Networks through Sparse Direct Feedback Alignment and Algorithm-Hardware Co-Design

Michigan

Agile and Miniature RF Front-ends Employing Multifunctional Ferroelectrics

Princeton

Towards Efficient and Private Deep Learning using 3-party Secure Computation

Rutgers

StructNN: Algorithm-Hardware Co-Design for Energy-Efficient Deep Neural Network using Structured Matrices

Stanford

Improving Communication Systems with Deep Generative Models

Stanford

Modern Error-Correcting Codes for Nanopore Sequencing-based Data Storage in DNA

Stanford

Next Generation Sensing Systems for Fully-Passive Remote Data Telemetry

Stanford

Efficient Reward Learning for Autonomous Driving and Robotics

Stanford

Learning Latent Structures Among HumanDrivers for Intent Inference and Planning

Stanford

Real-World Autonomous Agents from Virtual Environment: Generalizable Simulation and Representation Learning

Stanford

Accelerating Reinforcement Learning with Sub-optimal Teachers

UCB

Generic Hyperdimensional Processor for efficient human-centric IoT

UCB

Compositional Security for SoC Design

UCLA

Massively-Parallel Convolutional Neural Network Acceleration through Stochastic Computing

UCLA

A Wideband Frequency-Channelized ADC Using Time-Varying Circuits and Adaptive Digital Control

UCLA

Opportunistic Learning: Budgeted Cost-Sensitive Learning from Data Streams

UCLA & UIUC

Revolutionizing Large-Scale Graph Processing using Waferscale Architecture

UCSB

Enabling the First High Voltage Complementary-MOS (CMOS) technology based on Gallium nitride(GaN) for next generation ultra-fast, high-power efficient switching applications

UCSB & USC

EdgeML: Fast, Robust and Secure Machine Learning at the Edge

UCSD

Acquiring Compact AI with Natural Evolution

UCSD

Artificial Neural and Optic-Neural Synaptic Device for Neuromorphic Computing

UCSD

Toward Printable Ubiquitous Internet of Things with Capacitive Sensing, Communications and Identification Tags

UCSD

SweepSense : Sensing 5Ghz in 5Milliseconds

UIUC

Plug&Play Agents: Collaborative Learning in New Environments

UMD

Reliability-aware circuit design for self-diagnosis and self-prognosis in deep submicron CMOS

USC

Collaborative Sensing and Cooperative Guidance for Autonomous Cars via Scalable Perception Augmentation

Wisconsin-Madison

RAVEN: A Reconfigurable Architecture for Varying Emerging Neural Networks

Wisconsin-Madison

In-Home Functional Impairment Assessment of Stroke Patients Using Ubiquitous Millimeter Wave Devices

Wisconsin-Madison

Reevaluating the GPU Pipeline for VR/AR Acceleration for Mobile Devices

The QIF 2019 Finals will be held at Qualcomm headquarters in San Diego on April 3rd and 4th, 2019.

The Finals presentations are a major part of the winner selection. Both members of each finalist team are expected to attend and present at the Finals.

About the Finals presentations

Prepare a 12 - minute presentation (with an additional 3 min for questions). Please practice to ensure you adhere to the time limit.

You must submit your presentation slides in PDF or PowerPoint format and poster in PDF only format by 10:00AM PST on Thursday March 28th, 2019.

The submission site will be open on March 21, 2019. Details of the format of the slides and poster are below. Use your assigned confirmation number for the submission.

SUBMISSION SHOULD BE A SINGLE ZIP FILE CONTAINING THE THREE OR FOUR DOCUMENTS: PRESENTATION SLIDES, PDF OF SLIDES _FORPRINT, FIRST POSTER, AND OPTIONAL SECOND POSTER.

The presentations you submit will be preloaded on a computer provided by us for the Finals. Due to the presentation setup and to minimize the switching time between teams, you cannot use your own laptop for the presentation.

Students must use the submitted PowerPoint or PDF slides for their presentation.

The presentation schedule of the QIF Finals will be provided at registration on the first day.

About the slides

12-slides maximum, including the cover page. The presentation slides generally include:

  • The idea
  • The differentiating factors from state of the art
  • The execution plan and strength of the team

We do not recommend adding slides with references. If you do, they would also count towards the 12-slide limit.

Please do not use animations purely to get around the 12-slide limit.

Please ensure that all external media is embedded in the PDF or PowerPoint file itself, e.g., videos. The PDF/PowerPoint file must be self-contained.

In addition to your presentation, please provide a for-print PDF (8.5”X11” landscape) of your presentation (max. 12 pages). The name of the for-print PDF should end with "_forprint.pdf". It should have the same content (sans animations) as your presentation.

About the poster

Due to the tight program, we offer a poster session, where you have the opportunity for in-depth talks with the judges and Qualcomm engineers.

Each team may submit up to two posters (PDF format required, 24"x36").

We will print the posters for you ahead of time and provide the easels at the poster sessions.

See FAQs for more details

Do you need further Information?

Please direct your questions to: innovation.fellowship@qualcomm.com

We received 110 abstracts this year. After a careful selection process, the following have been chosen to advance to the next phase of the QIF 2019. Congratulations!

School Innovation Title
CalTech Theory and algorithms for remote tracking and control under data rate constraints
CMU Do-­it-­Yourself-­Locally: An IoT architecture For Localized Data Control for Privacy and Security
CMU Geometry-Aware Recurrent Networks: A Visual System for Embodied Agents
CMU Exocoustics: Real-time plug-and-play acoustic activity recognition in the wild
CMU Robust and Scalable Machine Learning on the Edge
CMU Improving Edge Intelligence with Distributed Inference
CMU MADS: A Machine Learning Assisted Diagnosis System
Cornell Xcel: An Architecture and Language for Application Accelerators
Cornell Exploiting Spatial Structure for Dynamic DNN Compression
Cornell PPAC: In-Memory Accelerator for Matrix-Vector Products
Georgia Tech Enabling Low-Cost Hardware Security against Side-Channels
Georgia Tech Hardware-Software Co-design for Deep Learning Acceleration
Georgia Tech Enabling Efficient Training of Deep Neural Networks through Sparse Deep Feedback Alignment and Algorithm-Hardware Co-Design
Georgia Tech A Wideband Mm-Wave Full-FoV MIMO Receiver Array for 5G MIMO in Complex, Dynamic Environments with Ultra-Low Latency Cross-Domain Space-Time Processing
Michigan InGaN channel HEMTs for high Power and mm-wave Applications
Michigan Fully Integrated, Millimeter-Wave, mm-Scale Imaging and Motion Sensor for 5G IoT
Michigan Towards Optimal Code Design for Non-Stationary Setups: From Wireless Channels to Distributed Computing Systems
Michigan Agile and Miniature RF Front-ends Employing Multifunctional Ferroelectrics
MIT Long-range low-power multi-modal wireless sensors
Princeton Towards Efficient and Private Deep Learning using 3-party Secure Computation
Princeton Securing Deep Neural Networks Against Out-of-Distribution Adversarial Examples
Rutgers StructNN: Algorithm-Hardware Co-Design for Energy-Efficient Deep Neural Network using Structured Matrices
Stanford Visual Question-Answering for 3D Data
Stanford Improving Communication Systems with Deep Generative Models
Stanford Modern Error-Correcting Codes for Nanopore Sequencing-based Data Storage in DNA
Stanford Next Generation Sensing Systems for Fully-Passive Remote Data Telemetry
Stanford Efficient Reward Learning for Autonomous Driving and Robotics
Stanford Ultra-Low Leakage Vertical DRAM Technology for Data-Intensive and Edge Computing
Stanford Learning Latent Structures Among HumanDrivers for Intent Inference and Planning
Stanford Real-World Autonomous Agents from Virtual Environment: Generalizable Simulation and Representation Learning
Stanford Accelerating Reinforcement Learning with Sub-optimal Teachers
UCB Generic Hyperdimensional Processor for efficient human-centric IoT
UCB Compositional Security for SoC Design
UCB Structurally Robust Neural Networks
UCLA Massively-Parallel Convolutional Neural Network Acceleration through Stochastic Computing
UCLA Interposer Breadboard to Democratize Chiplet-Based Design Ecosystem
UCLA A Wideband Frequency-Channelized ADC Using Time-Varying Circuits and Adaptive Digital Control
UCLA A Dynamical Chip-Scale LiDAR Solution for Environmental Mapping
UCLA Opportunistic Learning: Budgeted Cost-Sensitive Learning from Data Streams
UCLA Making Analog Deep Neural Networks Practicable
UCLA Good semi-supervised learning that requires both good and bad GAN
UCLA Parametrically Quality-Factor-Enhanced Surface Acoustic Wave (SAW) Filters for Mobile Applications
UCLA & UIUC Revolutionizing Large-Scale Graph Processing using Waferscale Architecture
UCSB Enabling the First High Voltage Complementary-MOS (CMOS) technology based on Gallium nitride(GaN) for next generation ultra-fast, high-power efficient switching applications
UCSB Towards High-Performance Reconfigurable RF Front-Ends with On-Chip Machine Learning
UCSB & USC EdgeML: Fast, Robust and Secure Machine Learning at the Edge
UCSD Acquiring Compact AI with Natural Evolution
UCSD Artificial Neural and Optic-Neural Synaptic Device for Neuromorphic Computing
UCSD PHD: Processing-in-Memory based HyperDimensional Computing for Light-Weight Learning
UCSD Toward Printable Ubiquitous Internet of Things with Capacitive Sensing, Communications and Identification Tags
UCSD SweepSense: Sensing 5Ghz in 5Milliseconds
UIUC A Reconfigurable 60 MHz-30 GHz PLL-less Ultra-Low Noise Frequency Synthesizer for Next Generation of IoT and Wireless Networks
UIUC Plug&Play Agents: Collaborative Learning in New Environments
UIUC Certified Adversarial Defenses via Semantic Transformations
UMD Enhancing IP Security of SoCs using Hardware Obfuscation
UMD Reliability-aware circuit design for self-diagnosis and self-prognosis in deep submicron technologies
UMD Systems Technology Integration for Biosensor Internet of Things Applications
University of Toronto PPG-Key management: A Highly Secured Encryption Method for Access Control
USC Collaborative Sensing and Cooperative Guidance for Autonomous Cars via Scalable Perception Augmentation
UT Austin Zero-static power radio-frequency switches based on 2D semiconductors
Washington Overlaying Occupied Spectrum with Downlink IoT Transmissions
Washington Automatically Generating Optimized Datatypes for Deep Learning
Wisconsin-Madison RAVEN: A Reconfigurable Architecture for Varying Emerging Neural Networks
Wisconsin-Madison In-Home Functional Impairment Assessment of Stroke Patients Using Ubiquitous Millimeter Wave Devices
Wisconsin-Madison Reevaluating the GPU Pipeline for VR/AR Acceleration for Mobile Devices

Qualcomm Research is a division of Qualcomm Technologies, Inc.

Qualcomm is inviting applications for the Qualcomm Innovation Fellowship 2019 from PhD students in the Electrical Engineering and Computer Science (and related) departments at:

United States

  • California Institute of Technology
  • Carnegie Mellon University (CMU)
  • Columbia University
  • Cornell University
  • Georgia Institute of Technology
  • Massachusetts Institute of Technology (MIT)
  • Princeton University
  • Rutgers University
  • Stanford University
  • UC Berkeley (UCB)
  • UC Los Angeles (UCLA)
  • UC San Diego (UCSD)
  • UC Santa Barbara (UCSB)
  • University of Illinois at Urbana-Champaign (UIUC)
  • University of Maryland, College Park
  • University of Michigan
  • University of Southern California (USC)
  • University of Texas at Austin
  • University of Washington
  • University of Wisconsin-Madison

Canada

  • University of Montreal
  • University of Toronto

We invite teams to submit proposals in the following areas. We also welcome proposals outside of the sub-areas listed below:

Advanced Semiconductor Electronics

  • Ultra-low (uW) power embedded platform for edge computing (ULP architectures and designs, HW accelerators, power generation and management, novel memories, security)
  • Novel materials and heterogeneous integration (2D semiconductors, GaAs, GaN, etc.)
  • CMOS package integration (thermal-aware designs or circuits, advanced packaging techniques, antenna-in-package etc.)
  • RF / analog ASICs and architectures (Sub-6GHz 5G power amplifiers, mmWave RFIC and Data Converters for 5G NR, adaptive RF signal processing algorithms, etc.)
  • Advanced antenna (millimeter-wave and phase-array antennas), novel antenna materials, structures and implementations
  • Power Management ASICs (wide bandwidth SMPS, wide bandwidth envelope tracker, embedded regulation)

Advances in Communication Techniques and Theory

  • Ultra-reliable and low latency communications
  • Wide-area wireless networks using high-frequency and mmWave spectrum
  • Massive MIMO, network MIMO, and coordinated multipoint processing
  • Wireless systems for unlicensed/shared spectrum
  • Low energy networks (Bluetooth LE, 802.15.4, Zigbee, Wi-Fi, etc.)
  • Advanced low power HW/FW/SW modem implementation approaches

Autonomous Driving

  • Advanced sensors and sensor fusion
  • Imaging radar
  • Computer vision for autonomy
  • Sensor fusion with deep learning
  • Behavior planning with uncertainty

Machine Learning

  • Natural language processing
  • Computer vision
  • Reinforcement and continual learning
  • On-device training
  • Intermediate representation for machine learning workloads/compilers
  • Transfer learning and Knowledge distillation
  • Novel compute architectures for ML tasks, e.g. in-memory compute, analog compute
  • Extreme energy efficient inference hardware accelerators for ML loads and lower complexity algorithms and convolutional nets
  • Generic Attribution methods for deriving non NN-based solutions and NN simplification

Multimedia Computing

  • Real Time 3D perception, mapping, reconstruction, and geometry interpretation
  • Eye-tracking devices and algorithms
  • Hand skeleton and multimodal human body pose estimation and tracking
  • Low power/complexity rendering systems
  • Lighting/illumination modelling
  • Multi-focal, near eye displays
  • High efficiency video coding techniques
  • Deep learning based image and video compression (intra and inter prediction, in-loop filters, transforms, entropy coding)
  • Deep learning based optimized video encoding
  • Perceptually optimized video coding
  • Image and video quality assessment
  • 6DoF video compression, Point Cloud compression

Processor Architecture and Implementation

  • Novel processor architectures, microarchitectures, extensions, and accelerators
  • Multimedia and gaming architectures (not limited to GPU, GPGPU, VLIW, DSP, etc.)
  • Novel architectures for artificial intelligence, edge training and inference
  • Security features of CPUs and accelerators at the instruction set, memory system, and SOC levels

Secure System Design

  • Isolation technologies: Virtualization, enclaves, and software sandboxing
  • Key management for IoT: Establishing trust between embedded devices
  • Machine learning model security: DRM for learning models
  • Protocol security: Analysis and verification of communication protocols
  • SoC security: Security of heterogeneous systems on chip
  • Software-based exploitation of hardware vulnerabilities: Micro-architectural attacks, side-channels, and associated countermeasures
  • User authentication: Biometric and behavioral authentication of users by mobile and embedded devices
  • Vulnerability detection: New tools and techniques for finding exploitable vulnerabilities in C software, with a focus on embedded systems

Semiconductor Test, Quality and Yield Learning

  • Defect-oriented testing and fault modeling in deep sub-micron process nodes
  • Applications of Data Analytics, Machine Learning and AI in Test
  • Test Challenges for 2.5D/3D Systems in Packages

Submission site opens: December 17, 2018

Application submission deadline: January 9, 2019 (10:00AM PST)

Selection announcement: January 22, 2019

Proposal submission deadline: February 12, 2019 (10:00AM PST)

Finalist presentations submission deadline: March 28, 2019 (10:00AM PST)

Finalist presentations: April 3 & 4, 2019

Winners’ announcement: May 29, 2019

QIF Winners’ Day: October 2019

Fellowship Winners and Finalists

The Qualcomm Innovation Fellowship began in 2009, and has continued to grow with the addition of more universities, more candidates, and expansion to our research centers internationally. Take a look at a list of all our fellowship winners and finalists from years past:

Qualcomm Innovation Fellowship Finalists' Day

May 14, 2018

1:35