mlsanityData to 7 Oct 2026

Lightweight compression of neural network feature tensors for collaborative intelligence

arXiv 2105.06002 ↗First posted 12 May 2021cs.LG eess.IVcited by 17

Not on the map. Why?

Abstract (arXiv)

In collaborative intelligence applications, part of a deep neural network (DNN) is deployed on a relatively low-complexity device such as a mobile phone or edge device, and the remainder of the DNN is processed where more computing resources are available, such as in the cloud. This paper presents a novel lightweight compression technique designed specifically to code the activations of a split DNN layer, while having a low complexity suitable for edge devices and not requiring any retraining. We also present a modified entropy-constrained quantizer design algorithm optimized for clipped activations. When applied to popular object-detection and classification DNNs, we were able to compress the 32-bit floating point activations down to 0.6 to 0.8 bits, while keeping the loss in accuracy to less than 1%. When compared to HEVC, we found that the lightweight codec consistently provided better inference accuracy, by up to 1.3%. The performance and simplicity of this lightweight compression technique makes it an attractive option for coding a layer's activations in split neural networks for edge/cloud applications.

Key facts. Cited by 17 papers in the map (data through 7 Oct 2026). Builds on 16 of its 33 references that are in the map. First posted on arXiv 12 May 2021.

Builds on

16 of 33 references.

References · 16

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2014Microsoft COCO: Common Objects in Contextcited by 17,498
  3. 2014ImageNet Large Scale Visual Recognition Challengecited by 8,441
  4. 2018YOLOv3: An Incremental Improvementcited by 2,652
  5. 2018MnasNet: Platform-Aware Neural Architecture Search for Mobilecited by 937
  6. 2016XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networkscited by 829
  7. 2018Quantizing deep convolutional networks for efficient inference: A whitepapercited by 507
  8. 2019Improving Neural Network Quantization without Retraining using Outlier Channel Splittingcited by 150
  9. 2018Scalable Methods for 8-bit Training of Neural Networkscited by 139
  10. 2019BottleNet: A Deep Learning Architecture for Intelligent Mobile Cloud Computing Servicescited by 67
  11. 2017WRPN: Wide Reduced-Precision Networkscited by 49
  12. 2018Deep feature compression for collaborative object detectioncited by 48
4 more
  1. 2018Post-training 4-bit quantization of convolution networks for rapid-deploymentcited by 38
  2. 2018Near-Lossless Deep Feature Compression for Collaborative Intelligencecited by 17
  3. 2020Back-and-Forth prediction for deep tensor compressioncited by 16
  4. 2019Towards Collaborative Intelligence Friendly Architectures for Deep Learningcited by 7

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 4; 2022: 2; 2023: 3; 2024: 3; 2025: 2; 2026: 3.

17 papers.

Sort:

By area:

  1. 2022SplitNets: Designing Neural Architectures for Efficient Distributed Computing on Head-Mounted Systemscited by 13
  2. 2021Lightweight Compression of Intermediate Neural Network Features for Collaborative Intelligencecited by 7
  3. 2022I-SPLIT: Deep Network Interpretability for Split Computingcited by 7
  4. 2021CALTeC: Content-Adaptive Linear Tensor Completion for Collaborative Intelligencecited by 5
  5. 2023Slimmable Encoders for Flexible Split DNNs in Bandwidth and Resource Constrained IoT Systemscited by 3
  6. 2024On the Impact of Black-box Deployment Strategies for Edge AI on Latency and Model Performancecited by 3
  7. 2025Memory- and Latency-Constrained Inference of Large Language Models via Adaptive Split Computingcited by 3
  8. 2021Revisit Visual Representation in Analytics Taxonomy: A Compression Perspectivecited by 1
  9. 2026AutoEncoder-Compressed Parallel Split Learning for Pre-trained Model Fine-Tuningcited by 1
  10. 2021DFTS2: Simulating Deep Feature Transmission Over Packet Loss Channelscited by 0
  11. 2023MetaMorphosis: Task-oriented Privacy Cognizant Feature Generation for Multi-task Learningcited by 0
  12. 2023Efficient IoT Inference via Context-Awarenesscited by 0
5 more
  1. 2024Inter-Feature-Map Differential Coding of Surveillance Videocited by 0
  2. 2024Aligned Vector Quantization for Edge-Cloud Collabrative Vision-Language Modelscited by 0
  3. 2025Range Asymmetric Numeral Systems-Based Lightweight Intermediate Feature Compression for Split Computing of Deep Neural Networkscited by 0
  4. 2026Bandwidth-adaptive Cloud-Assisted 360-Degree 3D Perception for Autonomous Vehiclescited by 0
  5. 2026Trainable Bitwise Soft Quantization for Input Feature Compressioncited by 0