Medusa: Universal Feature Learning via Attentional Multitasking
Not on the map. Why?
Abstract (arXiv)
Recent approaches to multi-task learning (MTL) have focused on modelling connections between tasks at the decoder level. This leads to a tight coupling between tasks, which need retraining if a new task is inserted or removed. We argue that MTL is a stepping stone towards universal feature learning (UFL), which is the ability to learn generic features that can be applied to new tasks without retraining. We propose Medusa to realize this goal, designing task heads with dual attention mechanisms. The shared feature attention masks relevant backbone features for each task, allowing it to learn a generic representation. Meanwhile, a novel Multi-Scale Attention head allows the network to better combine per-task features from different scales when making the final prediction. We show the effectiveness of Medusa in UFL (+13.18% improvement), while maintaining MTL performance and being 25% more efficient than previous approaches.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 27 of its 52 references that are in the map. First posted on arXiv 12 Apr 2022.
Builds on
27 of 52 references.
References · 27
- 2016DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFscited by 3,239
- 2018Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentationcited by 3,210
- 2019Deep High-Resolution Representation Learning for Human Pose Estimationcited by 1,339
- 2017SuperPoint: Self-Supervised Interest Point Detection and Descriptioncited by 1,313
- 2019A Comprehensive Survey on Transfer Learningcited by 1,154
- 2017An Overview of Multi-Task Learning in Deep Neural Networkscited by 791
- 2016Language Modeling with Gated Convolutional Networkscited by 688
- 2018Taskonomy: Disentangling Task Transfer Learningcited by 631
- 2018End-to-End Multi-Task Learning with Attentioncited by 533
- 2016Cross-stitch Networks for Multi-task Learningcited by 466
- 2018A Survey on Deep Transfer Learningcited by 417
- 2019D2-Net: A Trainable CNN for Joint Detection and Description of Local Featurescited by 324
15 more
- 2014Detect What You Can: Detecting and Representing Objects using Holistic Models and Body Partscited by 230
- 2018PAD-Net: Multi-Tasks Guided Prediction-and-Distillation Network for Simultaneous Depth Estimation and Scene Parsingcited by 224
- 2016UberNet: Training a `Universal' Convolutional Neural Network for Low-, Mid-, and High-Level Vision using Diverse Datasets and Limited Memorycited by 200
- 2020MTI-Net: Multi-Scale Task Interaction Networks for Multi-Task Learningcited by 157
- 2019Attentive Single-Tasking of Multiple Taskscited by 148
- 2019Pattern-Affinitive Propagation across Depth, Surface Normal and Semantic Segmentationcited by 141
- 2018NDDR-CNN: Layerwise Feature Fusing in Multi-Task CNNs by Neural Discriminative Dimensionality Reductioncited by 127
- 2020Learning to Branch for Multi-Task Learningcited by 114
- 2016Fully-adaptive Feature Sharing in Multi-Task Networks with Applications in Person Attribute Classificationcited by 93
- 2019Representation Similarity Analysis for Efficient Task taxonomy & Transfer Learningcited by 86
- 2019Branched Multi-Task Networks: Deciding What Layers To Sharecited by 76
- 2020Automated Search for Resource-Efficient Branched Multi-Task Networkscited by 51
- 2018A Modulation Module for Multi-task Learning with Applications in Image Retrievalcited by 49
- 2017PixelNet: Representation of the pixels, by the pixels, and for the pixelscited by 17
- 2019Scale-Adaptive Neural Dense Features: Learning via Hierarchical Context Aggregationcited by 5
Built on by
Citing papers per year: 2023: 1.
1 paper.