skip to main content

Title: Two Routes to Scalable Credit Assignment without Weight Symmetry, International Conference on Machine
The neural plausibility of backpropagation has long been disputed, primarily for its use of non-local weight transport — the biologically dubious requirement that one neuron instantaneously measure the synaptic weights of another. Until recently, attempts to create local learning rules that avoid weight transport have typically failed in the large-scale learning scenarios where backpropagation shines, e.g. ImageNet categorization with deep convolutional networks. Here, we investigate a recently proposed local learning rule that yields competitive performance with backpropagation and find that it is highly sensitive to metaparameter choices, requiring laborious tuning that does not transfer across network architecture. Our analysis indicates the underlying mathematical reason for this instability, allowing us to identify a more robust local learning rule that better transfers without metaparameter tuning. Nonetheless, we find a performance and stability gap between this local rule and backpropagation that widens with increasing model depth. We then investigate several non-local learning rules that relax the need for instantaneous weight transport into a more biologically-plausible "weight estimation" process, showing that these rules match state-of-the-art performance on deep networks and operate effectively in the presence of noisy updates. Taken together, our results suggest two routes towards the discovery of neural implementations for credit assignment more » without weight symmetry: further improvement of local rules so that they perform consistently across architectures and the identification of biological implementations for non-local learning mechanisms. « less
; ; ; ; ;
Award ID(s):
Publication Date:
Journal Name:
Proceedings of Machine Learning Research
Sponsoring Org:
National Science Foundation
More Like this
  1. Asynchronous event-driven computation and communication using spikes facilitate the realization of spiking neural networks (SNN) to be massively parallel, extremely energy efficient and highly robust on specialized neuromorphic hardware. However, the lack of a unified robust learning algorithm limits the SNN to shallow networks with low accuracies. Artificial neural networks (ANN), however, have the backpropagation algorithm which can utilize gradient descent to train networks which are locally robust universal function approximators. But backpropagation algorithm is neither biologically plausible nor neuromorphic implementation friendly because it requires: 1) separate backward and forward passes, 2) differentiable neurons, 3) high-precision propagated errors, 4) coherent copy of weight matrices at feedforward weights and the backward pass, and 5) non-local weight update. Thus, we propose an approximation of the backpropagation algorithm completely with spiking neurons and extend it to a local weight update rule which resembles a biologically plausible learning rule spike-timing-dependent plasticity (STDP). This will enable error propagation through spiking neurons for a more biologically plausible and neuromorphic implementation friendly backpropagation algorithm for SNNs. We test the proposed algorithm on various traditional and non-traditional benchmarks with competitive results.
  2. The brain modifies its synaptic strengths during learning in order to better adapt to its environment. However, the underlying plasticity rules that govern learning are unknown. Many proposals have been suggested, including Hebbian mechanisms, explicit error backpropagation, and a variety of alternatives. It is an open question as to what specific experimental measurements would need to be made to determine whether any given learning rule is operative in a real biological system. In this work, we take a "virtual experimental" approach to this problem. Simulating idealized neuroscience experiments with artificial neural networks, we generate a large-scale dataset of learning trajectories of aggregate statistics measured in a variety of neural network architectures, loss functions, learning rule hyperparameters, and parameter initializations. We then take a discriminative approach, training linear and simple non-linear classifiers to identify learning rules from features based on these observables. We show that different classes of learning rules can be separated solely on the basis of aggregate statistics of the weights, activations, or instantaneous layer-wise activity changes, and that these results generalize to limited access to the trajectory and held-out architectures and learning curricula. We identify the statistics of each observable that are most relevant for rule identification, findingmore »that statistics from network activities across training are more robust to unit undersampling and measurement noise than those obtained from the synaptic strengths. Our results suggest that activation patterns, available from electrophysiological recordings of post-synaptic activities on the order of several hundred units, frequently measured at wider intervals over the course of learning, may provide a good basis on which to identify learning rules.« less
  3. In many real-world applications, fully-differentiable RNNs such as LSTMs and GRUs have been widely deployed to solve time series learning tasks. These networks train via Backpropagation Through Time, which can work well in practice but involves a biologically unrealistic unrolling of the network in time for gradient updates, are computationally expensive, and can be hard to tune. A second paradigm, Reservoir Computing, keeps the recurrent weight matrix fixed and random. Here, we propose a novel hybrid network, which we call Hybrid Backpropagation Parallel Echo State Network (HBP-ESN) which combines the effectiveness of learning random temporal features of reservoirs with the readout power of a deep neural network with batch normalization. We demonstrate that our new network outperforms LSTMs and GRUs, including multi-layer "deep" versions of these networks, on two complex real-world multi-dimensional time series datasets: gesture recognition using skeleton keypoints from ChaLearn, and the DEAP dataset for emotion recognition from EEG measurements. We show also that the inclusion of a novel meta-ring structure, which we call HBP-ESN M-Ring, achieves similar performance to one large reservoir while decreasing the memory required by an order of magnitude. We thus offer this new hybrid reservoir deep learning paradigm as a new alternative directionmore »for RNN learning of temporal or sequential data.« less
  4. We use a recently developed synchronous Spiking Neural Network (SNN) model to study the problem of learning hierarchically-structured concepts. We introduce an abstract data model that describes simple hierarchical concepts. We define a feed-forward layered SNN model, with learning modeled using Oja’s local learning rule, a well known biologically-plausible rule for adjusting synapse weights. We define what it means for such a network to recognize hierarchical concepts; our notion of recognition is robust, in that it tolerates a bounded amount of noise. Then, we present a learning algorithm by which a layered network may learn to recognize hierarchical concepts according to our robust definition. We analyze correctness and performance rigorously; the amount of time required to learn each concept, after learning all of the sub-concepts, is approximately O ( 1ηk(`max log(k) + 1ε) + b log(k)), where k is the number of sub-concepts per concept, `max is the maximum hierarchical depth, η is the learning rate, ε describes the amount of uncertainty allowed in robust recognition, and b describes the amount of weight decrease for "irrelevant" edges. An interesting feature of this algorithm is that it allows the network to learn sub-concepts in a highly interleaved manner. This algorithm assumesmore »that the concepts are presented in a noise-free way; we also extend these results to accommodate noise in the learning process. Finally, we give a simple lower bound saying that, in order to recognize concepts with hierarchical depth two with noise-tolerance, a neural network should have at least two layers. The results in this paper represent first steps in the theoretical study of hierarchical concepts using SNNs. The cases studied here are basic, but they suggest many directions for extensions to more elaborate and realistic cases.« less
  5. Spiking neural networks (SNNs) well support spatio-temporal learning and energy-efficient event-driven hardware neuromorphic processors. As an important class of SNNs, recurrent spiking neural networks (RSNNs) possess great computational power. However, the practical application of RSNNs is severely limited by challenges in training. Biologically-inspired unsupervised learning has limited capability in boosting the performance of RSNNs. On the other hand, existing backpropagation (BP) methods suffer from high complexity of unfolding in time, vanishing and exploding gradients, and approximate differentiation of discontinuous spiking activities when applied to RSNNs. To enable supervised training of RSNNs under a well-defined loss function, we present a novel Spike-Train level RSNNs Backpropagation (ST-RSBP) algorithm for training deep RSNNs. The proposed ST-RSBP directly computes the gradient of a rate-coded loss function defined at the output layer of the network w.r.t tunable parameters. The scalability of ST-RSBP is achieved by the proposed spike-train level computation during which temporal effects of the SNN is captured in both the forward and backward pass of BP. Our ST-RSBP algorithm can be broadly applied to RSNNs with a single recurrent layer or deep RSNNs with multiple feedforward and recurrent layers. Based upon challenging speech and image datasets including TI46, N-TIDIGITS, Fashion-MNIST and MNIST, ST-RSBPmore »is able to train SNNs with an accuracy surpassing that of the current state-of-the-art SNN BP algorithms and conventional non-spiking deep learning models.« less