Dynamic Task Prioritization for Multitask Learning

By Michelle Guo, Albert Haque, De-An Huang, Serena Yeung, Li Fei-Fei. 2018

457 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

Solving Rubik's Cube with a Robot Hand (2019, methodology): “To balance these many losses, which lie on different scales, we use focal loss weighting as described in [37] to dynamically and automatically assign loss weights.”
Reparameterizing Convolutions for Incremental Multi-Task Learning without Task Interference (2020, background): “ayer sharing, the approaches above typically suer from task interference. Several works proposed to dynamically re-weight the loss function of each task [6,21,59,56], sort the order of task learning [16], or adapt the feature sharing between ‘related’ and ‘unrelated’ tasks [69], to mitigate the eect”
MultiGrain: a unified image embedding for classes and instances (2019, background): “…sharing of parameters [36], and ﬁnding the right optimization parameters for such networks in order to depart from the traditional setting of single-task single-dataset end-to-end gradient descent, and efﬁciently weight the gradients in order to obtain a well-performing network in all tasks [13].”
Ray Interference: a Source of Plateaus in Deep Reinforcement Learning (2019, background): “It is not uncommon to observe task dominance in this scenario (learning one task at a time [12]), and our analysis suggests possible reasons why tasks are sometimes learned sequentially despite the setup presenting them to the learning system all at once.”
Conditionally Adaptive Multi-Task Learning: Improving Transfer Learning in NLP Using Fewer Parameters & Less Data (2020, background): “As also discovered in MTL vision work (Guo et al., 2018), this type of prioritization on more difﬁcult tasks may explain MT-Uncertainty’s improved performance over other task selection methods.”
FairMOT: On the Fairness of Detection and Re-identification in Multiple Object Tracking (2020, unknown): “Multi-task Learning There is a large body of literature (Liu et al. 2019; Kendall et al. 2018; Chen et al. 2018b; Guo et al. 2018; Sener and Koltun 2018) on multi-task learning which may be used to balance the object detection and reID feature extraction tasks.”
Controlled Text Generation as Continuous Optimization with Multiple Constraints (2021, background): “Although this weighted summation formulation is intuitively appealing, it typically requires an expensive grid-search over the various scalings or use of a heuristic [25, 6, 18].”
Imbalance Problems in Object Detection: A Review (2019, methodology): “consistent and balanced optimization of the tasks. (iii) The difﬁculties of the tasks can be different, which affects the pace with which the tasks are learned, and hence hinders the training process [121]. Figure 14 illustrates a case where the loss of the classiﬁ- cation dominates the overall grad”
Multi-Task Learning for Dense Prediction Tasks: A Survey (2020, methodology): “This motivated researchers [8], [19], [20], [21] to balance the gradient magnitudes by setting the task-specific weights wi in the loss.”
Branched Multi-Task Networks: Deciding what layers to share (2019, background): “1In this setting, the input is first encoded through a stack of shared layers, after which tasks branch out into their own sequence of task-specific layers [14, 20, 21, 31, 43].”
Efficiently Identifying Task Groupings for Multi-Task Learning (2021, methodology): “In particular, dynamic loss reweighing has achieved performance superior to using fixed loss weights found with extensive hyperparameter search [28, 18, 37, 10, 45, 33].”
Multi-Task Learning with Deep Neural Networks: A Survey (2020, methodology): “To our knowledge, the only such works are Dynamic Task Prioritization (Guo et al., 2018) and the implicit schedule in (Jean et al., 2019).”
Pareto Multi-Task Learning (2019, background): “Recently, some learning-based methods have been proposed to solve MTL problems [36, 44].”
AuxNet: Auxiliary tasks enhanced Semantic Segmentation for Automated Driving (2019, background): “(Guo et al., 2018) on other hand proposes dynamic weight adjustments based on task difficulty.”
On Generating Extended Summaries of Long Documents (2020, unknown): “For example, the improved multi-tasking methods can involve task prioritization (Guo et al. 2018) to dynamically balance the learning process between two tasks during training, rather than using a fixedα parameter.”
Attentive Single-Tasking of Multiple Tasks (2019, methodology): “…according to each task’s uncertainty, while Sener and Koltun [56] estimate an adaptive weighting of the different task losses based on a pareto-optimal formulation of MTL. Similarly, [20] provide a MTL framework where tasks are dynamically sorted by difﬁculty and the hardest are learned ﬁrst.”
End-To-End Multi-Task Learning With Attention (2018, methodology): “As an alternative to using task losses to determine task difficulties, Dynamic Task Prioritisation [10] encourages prioritisation of difficult tasks directly using performance metrics such as accuracy and precision.”
NeurAll: Towards a Unified Model for Visual Perception in Automated Driving (2019, background): “Guo et al. [37] updated tasks weights by observing difﬁculty of the tasks during training.”
Conflict-Averse Gradient Descent for Multi-task Learning (2021, background): “, uncertainty [15], gradient norm [3], or difficulty [9].”