An Adaptive Random Path Selection Approach for Incremental Learning.

By Jathushan Rajasegaran, Munawar Hayat, Salman Hameed Khan, Fahad Shahbaz Khan, Ling Shao, Ming-Hsuan Yang. 2019

26 citations recorded by the source corpus.

In a conventional supervised learning setting, a machine learning model has access to examples of all object classes that are desired to be recognized during the inference stage. This results in a fixed model that lacks the flexibility to adapt to new learning tasks. In practical settings, learning tasks often arrive in a sequence and the models must continually learn to increment their previously acquired knowledge. Existing incremental learning approaches fall well below the state-of-the-art cumulative models that use all training classes at once. In this paper, we propose a random path selection algorithm, called Adaptive RPS-Net, that progressively chooses optimal paths for the new tasks while encouraging parameter sharing between tasks. We introduce a new network capacity measure that enables us to automatically switch paths if the already used resources are saturated. Since the proposed path-reuse strategy ensures forward knowledge transfer, our approach is efficient and has considerably less computation overhead. As an added novelty, the proposed model integrates knowledge distillation and retrospection along with the path selection strategy to overcome catastrophic forgetting. In order to maintain an equilibrium between previous and newly acquired knowledge, we propose a simple controller to dynamically balance the model plasticity. Through extensive experiments, we demonstrate that the Adaptive RPS-Net method surpasses the state-of-the-art performance for incremental learning and by utilizing parallel computation this method can run in constant time with nearly the same efficiency as a conventional deep convolutional neural network.

Subfields: Artificial Intelligence, Optimization & Training, Meta & Continual Learning

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

In Optimization & Training

In Meta & Continual Learning

Citations in the corpus

Learnability and Algorithm for Continual Learning (2023, methodology): “…buffer data (Rusu et al., 2016; LopezPaz & Ranzato, 2017; Chaudhry et al., 2019a; Hou et al., 2019; Wu et al., 2019; Rolnick et al., 2019; Buzzega et al., 2020; Rajasegaran et al., 2020a; Liu et al., 2021; Cha et al., 2021; Yan et al., 2021; Wang et al., 2022b; Guo et al., 2022; Kim et al., 2022a).”
A Theoretical Study on Solving Continual Learning (2022, background): “[35] Jathushan Rajasegaran, Munawar Hayat, Salman Khan, Fahad Shahbaz Khan, Ling Shao, and Ming-Hsuan Yang.”
Novel Class Discovery without Forgetting (2022, background): “Dynamically expanding and parameter isolation methods [44,40,39,1,33] form a third class of methods to address forgetting.”
Energy-based Latent Aligner for Incremental Learning (2022, background): “, by new network blocks or gating mechanisms [1, 31, 38, 39, 44].”
DyTox: Transformers for Continual Learning with DYnamic TOken eXpansion (2021, methodology): “RPSNet [21] uses knowledge distillation and manipulates subnetworks in its architecture, following the lottery ticket hypothesis [8].”
iTAML: An Incremental Task-Agnostic Meta-learning Approach (2020, methodology): “Existing IL methods propose architectural modiﬁcations for deep CNNs e.g., dynamic networks [21, 22], dual-memory modules [7], and network expansion [25].”
Continual Learning Based on OOD Detection and Task Masking (2022, methodology): “For A-RPS, CCG, and Co2L, we copy the results from their original papers as their codes are not released to the public or the public code cannot run on our system.”