Lifelong Learning of Spatiotemporal Representations With Dual-Memory Recurrent Self-Organization

By German Ignacio Parisi, Jun Tani, Cornelius Weber, Stefan Wermter. 2018

134 citations recorded by the source corpus.

Artificial autonomous agents and robots interacting in complex environments are required to continually acquire and fine-tune knowledge over sustained periods of time. The ability to learn from continuous streams of information is referred to as lifelong learning and represents a long-standing challenge for neural network models due to catastrophic forgetting in which novel sensory experience interferes with existing representations and leads to abrupt decreases in the performance on previously acquired knowledge. Computational models of lifelong learning typically alleviate catastrophic forgetting in experimental scenarios with given datasets of static images and limited complexity, thereby differing significantly from the conditions artificial agents are exposed to. In more natural settings, sequential information may become progressively available over time and access to previous experience may be restricted. Therefore, specialized neural network mechanisms are required that adapt to novel sequential experience while preventing disruptive interference with existing representations. In this paper, we propose a dual-memory self-organizing architecture for lifelong learning scenarios. The architecture comprises two growing recurrent networks with the complementary tasks of learning object instances (episodic memory) and categories (semantic memory). Both growing networks can expand in response to novel sensory experience: the episodic memory learns fine-grained spatiotemporal representations of object instances in an unsupervised fashion while the semantic memory uses task-relevant signals to regulate structural plasticity levels and develop more compact representations from episodic experience. For the consolidation of knowledge in the absence of external sensory input, the episodic memory periodically replays trajectories of neural reactivations. We evaluate the proposed model on the CORe50 benchmark dataset for continuous object recognition, showing that we significantly outperform current methods of lifelong learning in three different incremental learning scenarios.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Model Architectures, 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 Computer Vision and Pattern Recognition

In Model Architectures

In Meta & Continual Learning

Citations in the corpus

On the role of neurogenesis in overcoming catastrophic forgetting (2018, methodology): “Proposed solutions for mitigating catastrophic interference in this domain have focused on regulating intrinsic levels of plasticity to protect acquired knowledge given a fixed number of neurons [5, 6, 13], dynamically allocating new neurons to accommodate novel dissimilar knowledge [7, 8, 16], and ”
Continuous Learning in Single-Incremental-Task Scenarios (2018, background): “In particular, [25] proposes an interesting 2-level self-organizing model built on the top of a convolutional feature extractor that is capable of exploiting temporal coherence in CORe50 videos and provides good results also with weak supervision.”
Adaptive Orthogonal Projection for Batch and Online Continual Learning (2022, background): “Many other papers use related approaches (Zenke, Poole, and Ganguli 2017; Fernando et al. 2017; Aljundi et al. 2017; Ritter, Botev, and Barber 2018; Xu and Zhu 2018; Kemker and Kanan 2018; Parisi et al. 2018; Ahn et al. 2019; Hu et al. 2021; Dhar et al. 2019; Adel, Zhao, and Turner 2020).”
Continual Learning by Using Information of Each Class Holistically (2021, background): “Many papers dealing with CF use a similar idea (Zenke, Poole, and Ganguli 2017; Fernando et al. 2017; Aljundi et al. 2017; Ritter, Botev, and Barber 2018; Xu and Zhu 2018; Kemker and Kanan 2018; Parisi et al. 2018; Dhar et al. 2019; Adel, Zhao, and Turner 2020).”
Replay in Deep Learning: Current Approaches and Missing Biological Elements (2021, positive): “…of supervised learning literature has focused on over-coming forgetting in feedforward or convolutional neural networks, there has also been work focused on using replay to mitigate forgetting in recurrent neural networks (Parisi, Tani, Weber, and Wermter, 2018; Sodhani, Chandar, & Bengio, 2020).”
Imbalanced Continual Learning with Partitioning Reservoir Sampling (2020, background): “Specifically, there have been many ventures into but not limited to single-label text classification [19], question answering [19], language instruction and translation [50], object detection [77,53], captioning [64] and even video representation learning [67,68].”
Lifelong Machine Learning with Deep Streaming Linear Discriminant Analysis (2019, methodology): “Similar to [41], another future direction could include the creation of an SLDA model that accounts for the temporal structure of video data in its update and inference procedures.”
Continual Learning for Robotics (2019, methodology): “The use of dual architecture is explicit in many bio-inspired approaches such as [48, 51, 115, 154, 70].”
Rehearsal-Free Continual Learning over Small Non-I.I.D. Batches (2019, neutral): “Another interesting technique, reporting good results on CORe50, is the Dual-Memory Recurrent Self-Organization proposed in [25]: however, results included in that work are not directly comparable with our achievements because the aforementioned approach also exploits the temporal dimension of CORe5”
Unsupervised Progressive Learning and the STAM Architecture (2019, neutral): “Finally, brain-inspired dual-memory systems have been proposed before for memory consolidation (e.g., [Parisi et al. , 2018; Shin et al. , 2017]).”
CILEA-NET: Curriculum-Based Incremental Learning Framework for Remote Sensing Image Classification (2021, background): “Adding more neurons as suggested in [6] can also primarily alleviate the issue.”
Reactive Exploration to Cope with Non-Stationarity in Lifelong Reinforcement Learning (2022, background): “Some approaches in lifelong learning address continual domain shifts via rehearsal or pseudo-rehearsal methods (Parisi et al., 2018; Hayes et al., 2019), the reduction of representational overlap (Serra et al.”
Class Incremental Learning for Video Action Classification (2021, methodology): “Each neuron has a weight wi and a context descriptor ci [22, 23], which represent the weights of the current frame and the previous frames.”
Self-Net: Lifelong Learning via Continual Self-Modeling (2018, methodology): “Some methods speciﬁcally use autoencoders for this task (Zhou et al., 2012; Riemer et al., 2017; Parisi et al., 2018b).”
Online Continual Learning for Embedded Devices (2022, background): “Continual machine learning systems have the ability to learn from ever-growing data streams (Parisi et al., 2018).”