Dataset Information

Meta-learning biologically plausible plasticity rules with random feedback pathways.

ABSTRACT: Backpropagation is widely used to train artificial neural networks, but its relationship to synaptic plasticity in the brain is unknown. Some biological models of backpropagation rely on feedback projections that are symmetric with feedforward connections, but experiments do not corroborate the existence of such symmetric backward connectivity. Random feedback alignment offers an alternative model in which errors are propagated backward through fixed, random backward connections. This approach successfully trains shallow models, but learns slowly and does not perform well with deeper models or online learning. In this study, we develop a meta-learning approach to discover interpretable, biologically plausible plasticity rules that improve online learning performance with fixed random feedback connections. The resulting plasticity rules show improved online training of deep models in the low data regime. Our results highlight the potential of meta-learning to discover effective, interpretable learning rules satisfying biological constraints.

SUBMITTER: Shervani-Tabar N

PROVIDER: S-EPMC10066328 | biostudies-literature | 2023 Mar

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Meta-learning biologically plausible plasticity rules with random feedback pathways.

Shervani-Tabar Navid N Rosenbaum Robert R

Nature communications 20230331 1

Backpropagation is widely used to train artificial neural networks, but its relationship to synaptic plasticity in the brain is unknown. Some biological models of backpropagation rely on feedback projections that are symmetric with feedforward connections, but experiments do not corroborate the existence of such symmetric backward connectivity. Random feedback alignment offers an alternative model in which errors are propagated backward through fixed, random backward connections. This approach s ...[more]

PMID: 37002222

Similar Datasets

Project description:We develop biologically plausible training mechanisms for self-supervised learning (SSL) in deep networks. Specifically, by biologically plausible training we mean (i) all updates of weights are based on current activities of pre-synaptic units and current, or activity retrieved from short term memory of post synaptic units, including at the top-most error computing layer, (ii) complex computations such as normalization, inner products and division are avoided, (iii) asymmetric connections between units, and (iv) most learning is carried out in an unsupervised manner. SSL with a contrastive loss satisfies the third condition as it does not require labeled data and it introduces robustness to observed perturbations of objects, which occur naturally as objects or observers move in 3D and with variable lighting over time. We propose a contrastive hinge based loss whose error involves simple local computations satisfying (ii), as opposed to the standard contrastive losses employed in the literature, which do not lend themselves easily to implementation in a network architecture due to complex computations involving ratios and inner products. Furthermore, we show that learning can be performed with one of two more plausible alternatives to backpropagation that satisfy conditions (i) and (ii). The first is difference target propagation (DTP), which trains network parameters using target-based local losses and employs a Hebbian learning rule, thus overcoming the biologically implausible symmetric weight problem in backpropagation. The second is layer-wise learning, where each layer is directly connected to a layer computing the loss error. The layers are either updated sequentially in a greedy fashion (GLL) or in random order (RLL), and each training stage involves a single hidden layer network. Backpropagation through one layer needed for each such network can either be altered with fixed random feedback weights (RF) or using updated random feedback weights (URF) as in Amity's study 2019. Both methods represent alternatives to the symmetric weight issue of backpropagation. By training convolutional neural networks (CNNs) with SSL and DTP, GLL or RLL, we find that our proposed framework achieves comparable performance to standard BP learning downstream linear classifier evaluation of the learned embeddings.

Project description:Recurrent spiking neural networks (RSNN) in the brain learn to perform a wide range of perceptual, cognitive and motor tasks very efficiently in terms of energy consumption and their training requires very few examples. This motivates the search for biologically inspired learning rules for RSNNs, aiming to improve our understanding of brain computation and the efficiency of artificial intelligence. Several spiking models and learning rules have been proposed, but it remains a challenge to design RSNNs whose learning relies on biologically plausible mechanisms and are capable of solving complex temporal tasks. In this paper, we derive a learning rule, local to the synapse, from a simple mathematical principle, the maximization of the likelihood for the network to solve a specific task. We propose a novel target-based learning scheme in which the learning rule derived from likelihood maximization is used to mimic a specific spatio-temporal spike pattern that encodes the solution to complex temporal tasks. This method makes the learning extremely rapid and precise, outperforming state of the art algorithms for RSNNs. While error-based approaches, (e.g. e-prop) trial after trial optimize the internal sequence of spikes in order to progressively minimize the MSE we assume that a signal randomly projected from an external origin (e.g. from other brain areas) directly defines the target sequence. This facilitates the learning procedure since the network is trained from the beginning to reproduce the desired internal sequence. We propose two versions of our learning rule: spike-dependent and voltage-dependent. We find that the latter provides remarkable benefits in terms of learning speed and robustness to noise. We demonstrate the capacity of our model to tackle several problems like learning multidimensional trajectories and solving the classical temporal XOR benchmark. Finally, we show that an online approximation of the gradient ascent, in addition to guaranteeing complete locality in time and space, allows learning after very few presentations of the target output. Our model can be applied to different types of biological neurons. The analytically derived plasticity learning rule is specific to each neuron model and can produce a theoretical prediction for experimental validation.

Project description:Often, when animals encounter an unexpected sensory event, they transition from executing a variety of movements to repeating the movement(s) that may have caused the event. According to a recent theory of action discovery (Redgrave and Gurney, 2006), repetition allows the animal to represent those movements, and the outcome, as an action for later recruitment. The transition from variation to repetition often follows a non-random, structured, pattern. While the structure of the pattern can be explained by sophisticated cognitive mechanisms, simpler mechanisms based on dopaminergic modulation of basal ganglia (BG) activity are thought to underlie action discovery (Redgrave and Gurney, 2006). In this paper we ask the question: can simple BG-mediated mechanisms account for a structured transition from variation to repetition, or are more sophisticated cognitive mechanisms always necessary? To address this question, we present a computational model of BG-mediated biasing of behavior. In our model, unlike most other models of BG function, the BG biases behavior through modulation of cortical response to excitation; many possible movements are represented by the cortical area; and excitation to the cortical area is topographically-organized. We subject the model to simple reaching tasks, inspired by behavioral studies, in which a location to which to reach must be selected. Locations within a target area elicit a reinforcement signal. A structured transition from variation to repetition emerges from simple BG-mediated biasing of cortical response to excitation. We show how the structured pattern influences behavior in simple and complicated tasks. We also present analyses that describe the structured transition from variation to repetition due to BG-mediated biasing and from biasing that would be expected from a type of cognitive biasing, allowing us to compare behavior resulting from these types of biasing and make connections with future behavioral experiments.

Dataset Information

Meta-learning biologically plausible plasticity rules with random feedback pathways.

Publications

Meta-learning biologically plausible plasticity rules with random feedback pathways.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets