Abstract Artificial neural networks are known to suffer from catastrophic forgetting: when learning multiple tasks sequentially, they perform well on the most recent task at the expense of previously learned tasks. In the brain, sleep is known to play an important role in incremental learning by replaying recent and old conflicting memory traces. Here we tested the hypothesis that implementing a sleep-like phase in artificial neural networks can protect old memories during new training and alleviate catastrophic forgetting. Sleep was implemented as off-line training with local unsupervised Hebbian plasticity rules and noisy input. In an incremental learning framework, sleep was able to recover old tasks that were otherwise forgotten. Previously learned memories were replayed spontaneously during sleep, forming unique representations for each class of inputs. Representational sparseness and neuronal activity corresponding to the old tasks increased while new task related activity decreased. The study suggests that spontaneous replay simulating sleep-like dynamics can alleviate catastrophic forgetting in artificial neural networks.
more »
« less
Reducing Catastrophic Forgetting With Associative Learning: A Lesson From Fruit Flies
Catastrophic forgetting remains an outstanding challenge in continual learning. Recently, methods inspired by the brain, such as continual representation learning and memory replay, have been used to combat catastrophic forgetting. Associative learning (retaining associations between inputs and outputs, even after good representations are learned) plays an important function in the brain; however, its role in continual learning has not been carefully studied. Here, we identified a two-layer neural circuit in the fruit fly olfactory system that performs continual associative learning between odors and their associated valences. In the first layer, inputs (odors) are encoded using sparse, high-dimensional representations, which reduces memory interference by activating nonoverlapping populations of neurons for different odors. In the second layer, only the synapses between odor-activated neurons and the odor’s associated output neuron are modified during learning; the rest of the weights are frozen to prevent unrelated memories from being overwritten. We prove theoretically that these two perceptron-like layers help reduce catastrophic forgetting compared to the original perceptron algorithm, under continual learning. We then show empirically on benchmark data sets that this simple and lightweight architecture outperforms other popular neural-inspired algorithms when also using a two-layer feedforward architecture. Overall, fruit flies evolved an efficient continual associative learning algorithm, and circuit mechanisms from neuroscience can be translated to improve machine computation.
more » « less- Award ID(s):
- 2211386
- PAR ID:
- 10517683
- Publisher / Repository:
- MIT Press
- Date Published:
- Journal Name:
- Neural Computation
- Volume:
- 35
- Issue:
- 11
- ISSN:
- 0899-7667
- Page Range / eLocation ID:
- 1797 to 1819
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
null (Ed.)People learn throughout life. However, incrementally updating conventional neural networks leads to catastrophic forgetting. A common remedy is replay, which is inspired by how the brain consolidates memory. Replay involves fine-tuning a network on a mixture of new and old instances. While there is neuroscientific evidence that the brain replays compressed memories, existing methods for convolutional networks replay raw images. Here, we propose REMIND, a brain-inspired approach that enables efficient replay with compressed representations. REMIND is trained in an online manner, meaning it learns one example at a time, which is closer to how humans learn. Under the same constraints, REMIND outperforms other methods for incremental class learning on the ImageNet ILSVRC-2012 dataset. We probe REMIND’s robustness to data ordering schemes known to induce catastrophic forgetting. We demonstrate REMIND’s generality by pioneering online learning for Visual Question Answering (VQA).more » « less
-
Supervised Continual learning involves updating a deep neural network (DNN) from an ever-growing stream of labeled data. While most work has focused on overcoming catastrophic forgetting, one of the major motivations behind continual learning is being able to efficiently update a network with new information, rather than retraining from scratch on the training dataset as it grows over time. Despite recent continual learning methods largely solving the catastrophic forgetting problem, there has been little attention paid to the efficiency of these algorithms. Here, we study recent methods for incremental class learning and illustrate that many are highly inefficient in terms of compute, memory, and storage. Some methods even require more compute than training from scratch! We argue that for continual learning to have real-world applicability, the research community cannot ignore the resources used by these algorithms. There is more to continual learning than mitigating catastrophic forgetting.more » « less
-
Real-time on-device continual learning is needed for new applications such as home robots, user personalization on smartphones, and augmented/virtual reality headsets. However, this setting poses unique challenges: embedded devices have limited memory and compute capacity and conventional machine learning models suffer from catastrophic forgetting when updated on non-stationary data streams. While several online continual learning models have been developed, their effectiveness for embedded applications has not been rigorously studied. In this paper, we first identify criteria that online continual learners must meet to effectively perform real-time, on-device learning. We then study the efficacy of several online continual learning methods when used with mobile neural networks. We measure their performance, memory usage, compute requirements, and ability to generalize to out-of-domain inputs.more » « less
-
Neurons in real brains are complex computational units, capable of input-specific damping, inter-trial memory, and context-dependent signal processing. Artificial neurons, on the other hand, are usually implemented as simple weighted sums. Here we explore if increasing the computational power of individual neurons can yield more powerful neural networks. Specifically, we introduce Deep Artificial Neurons (DANs)—small neural networks with shared, learnable parameters embedded within a larger network. DANs act as filters between nodes in the net-work; namely, they receive vectorized inputs from multiple neurons in the previous layer, condense these signals into a single output, then send this processed signal to the neurons in the subsequent layer. We demonstrate that it is possible to meta-learn shared parameters for the various DANS in the network in order to facilitate continual and transfer learning during deployment. Specifically, we present experimental results on (1) incremental non-linear regression tasks and (2)unsupervised class-incremental image reconstruction that show that DANs allow a single network to update its synapses (i.e., regular weights) over time with minimal forgetting. Notably, our approach uses standard backpropagation, does not require experience replay, and does need separate wake/sleep phases.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1162/neco_a_01615
