At Qualcomm AI Research, we are advancing AI to make its core capabilities – perception, reasoning, and action – ubiquitous across devices. Our mission is to make breakthroughs in fundamental AI research and scale them across industries. By bringing together some of the best minds in the field, we’re pushing the boundaries of what’s possible and shaping the future of AI.
Qualcomm AI Research Introduction
Jun 8, 2018
2:27
“Machine learning algorithms such as deep learning already use large amounts of energy. And with AI increasingly moving to power-constrained edge devices, energy efficiency will only become more important. The benchmark that matters will be how much intelligence one can squeeze out of every joule of energy. I believe that Qualcomm is uniquely positioned to address this problem and be a key player in AI.”
- Professor Dr. Max Welling
VP of Technology, Qualcomm Technologies Netherlands B.V.
Model design, compression, quantization, activation, algorithms, and efficient hardware.
Continuous learning, model adaptation, and privacy-preserved distributed learning.
Robust learning through minimal data, unsupervised learning, and on-device learning.
Multi-task and multi-modal learning, sensor fusion, and cloud-edge systems.
Webinar: Pushing the boundaries of AI research
Sep 9, 2020
1:00:23
Webinar: Enabling power-efficient AI through quantization
May 1, 2020
1:01:18
Webinar: 5G+AI: The Ingredients Fueling Tomorrow's Technology Innovations
Feb 19, 2020
58:52
AI Research Video
Dec 6, 2019
2:46
Novel papers are one of the ways Qualcomm Technologies contributes impactful research to the larger community of AI research. Below are papers that Qualcomm AI Research has written or co-authored.
Lossy Compression with Distortion Constrained Optimization | CVPR 2020
Adversarial Distortion for Learned Video Compression | CVPR 2020
ActionBytes: Learning from Trimmed Videos to Localize Actions | CVPR 2020
End-to-End Lane Marker Detection via Row-wise Classification | CVPR 2020
A Data and Compute Efficient Design for Limited-Resources Deep Learning | ICLR 2020
Combinatorial Bayesian Optimization using Graph Cartesian Product | NeurIPS 2019
A General Theory of Equivariant CNNs on Homogeneous Spaces | NeurIPS 2019
Recognizing Compressed Videos: Challenges and Promises | ICCV 2019
Weakly-supervised Degree of Eye-closeness Estimation | ICCV 2019
Video Compression With Rate-Distortion Autoencoders | ICCV 2019
Video2vec Embeddings Recognize Events when Examples are Scarce | IEEE 2017
Gauge Equivariant Convolutional Networks and the Icosahedral CNN | ICML 2019
Batch-Shaping for Learning Conditional Channel Gated Networks" | ICLR 2020
Gradient l1 Regularization for Quantization Robustness | ICLR 2020
LSQ+: Improving low-bit quantization through learnable offsets and better initialization | CVPR 2020
Data-Free Quantization through Weight Equalization and Bias Correction | ICCV 2019
A Quantization-Friendly Separable Convolution for MobileNets | EMC^2 2018
Relaxed Quantization for Discretized Neural Networks | ICLR 2019
Understanding Straight-Through Estimator in Training Activation Quantized Neural Nets | ICLR 2019
DAC: Data-free Automatic Acceleration of Convolutional Networks | WACV 2019
Qualcomm Innovation Center open sourced AIMET, which includes state-of-the-art quantization and compression techniques. The goal for this open source project is to collaborate with other leading AI researchers, provide a simple library plugin for AI developers, and help migrate the ecosystem toward integer inference.
Are you a machine learning specialist eager to start a new challenge and have a global impact on the industry? Come join us as we make AI breakthroughs.
Qualcomm AI Research is an initiative of Qualcomm Technologies, Inc.
AI Model Efficiency Toolkit is a product of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a wholly-owned subsidiary of Qualcomm Technologies, Inc.
©2020 Qualcomm Technologies, Inc. and/or its affiliated companies.