arXiv subjects

Dominique Chu

Publications and source records attributed to Dominique Chu.

13 recordsLinked to original sources

arXiv2023

Random Feedback Alignment Algorithms to train Neural Networks: Why do they Align?

Feedback alignment algorithms are an alternative to backpropagation to train neural networks, whereby some of the partial derivatives that are required to compute the gradient are replaced by random terms. This essentially transforms the update rule into a random walk in weight space. Surprisingly, learning still works with those algorithms, including training of deep neural networks. This is generally attributed to an alignment of the update of the random walker with the true gradient - the eponymous gradient alignment -- which drives an approximate gradient descend. The mechanism that leads to this alignment remains unclear, however. In this paper, we use mathematical reasoning and simulations to investigate gradient alignment. We observe that the feedback alignment update rule has fixed points, which correspond to extrema of the loss function. We show that gradient alignment is a stability criterion for those fixed points. It is only a necessary criterion for algorithm performance. Experimentally, we demonstrate that high levels of gradient alignment can lead to poor algorithm performance and that the alignment is not always driving the gradient descend.

cs.LG↗

arXiv2022

Exploring Tradeoffs in Spiking Neural Networks

Spiking Neural Networks (SNNs) have emerged as a promising alternative to traditional Deep Neural Networks for low-power computing. However, the effectiveness of SNNs is not solely determined by their performance but also by their energy consumption, prediction speed, and robustness to noise. The recent method Fast \& Deep, along with others, achieves fast and energy-efficient computation by constraining neurons to fire at most once. Known as Time-To-First-Spike (TTFS), this constraint however restricts the capabilities of SNNs in many aspects. In this work, we explore the relationships between performance, energy consumption, speed and stability when using this constraint. More precisely, we highlight the existence of tradeoffs where performance and robustness are gained at the cost of sparsity and prediction latency. To improve these tradeoffs, we propose a relaxed version of Fast \& Deep that allows for multiple spikes per neuron. Our experiments show that relaxing the spike constraint provides higher performance while also benefiting from faster convergence, similar sparsity, comparable prediction latency, and better robustness to noise compared to TTFS SNNs. By highlighting the limitations of TTFS and demonstrating the advantages of unconstrained SNNs we provide valuable insight for the development of effective learning strategies for neuromorphic computing.

cs.NE↗

arXiv2022

Low-Variance Forward Gradients using Direct Feedback Alignment and Momentum

Supervised learning in deep neural networks is commonly performed using error backpropagation. However, the sequential propagation of errors during the backward pass limits its scalability and applicability to low-powered neuromorphic hardware. Therefore, there is growing interest in finding local alternatives to backpropagation. Recently proposed methods based on forward-mode automatic differentiation suffer from high variance in large deep neural networks, which affects convergence. In this paper, we propose the Forward Direct Feedback Alignment algorithm that combines Activity-Perturbed Forward Gradients with Direct Feedback Alignment and momentum. We provide both theoretical proofs and empirical evidence that our proposed method achieves lower variance than forward gradient techniques. In this way, our approach enables faster convergence and better performance when compared to other local alternatives to backpropagation and opens a new perspective for the development of online learning algorithms compatible with neuromorphic systems.

cs.LG↗

arXiv2022

Programming molecular systems to emulate a learning spiking neuron

Hebbian theory seeks to explain how the neurons in the brain adapt to stimuli, to enable learning. An interesting feature of Hebbian learning is that it is an unsupervised method and as such, does not require feedback, making it suitable in contexts where systems have to learn autonomously. This paper explores how molecular systems can be designed to show such proto-intelligent behaviours, and proposes the first chemical reaction network (CRN) that can exhibit autonomous Hebbian learning across arbitrarily many input channels. The system emulates a spiking neuron, and we demonstrate that it can learn statistical biases of incoming inputs. The basic CRN is a minimal, thermodynamically plausible set of micro-reversible chemical equations that can be analysed with respect to their energy requirements. However, to explore how such chemical systems might be engineered de novo, we also propose an extended version based on enzyme-driven compartmentalised reactions. Finally, we also show how a purely DNA system, built upon the paradigm of DNA strand displacement, can realise neuronal dynamics. Our analysis provides a compelling blueprint for exploring autonomous learning in biological settings, bringing us closer to realising real synthetic biological intelligence.

q-bio.NC↗

arXiv2021

Integrate-and-Fire Neurons for Low-Powered Pattern Recognition

Embedded systems acquire information about the real world from sensors and process it to make decisions and/or for transmission. In some situations, the relationship between the data and the decision is complex and/or the amount of data to transmit is large (e.g. in biologgers). Artificial Neural Networks (ANNs) can efficiently detect patterns in the input data which makes them suitable for decision making or compression of information for data transmission. However, ANNs require a substantial amount of energy which reduces the lifetime of battery-powered devices. Therefore, the use of Spiking Neural Networks can improve such systems by providing a way to efficiently process sensory data without being too energy-consuming. In this work, we introduce a low-powered neuron model called Integrate-and-Fire which exploits the charge and discharge properties of the capacitor. Using parallel and series RC circuits, we developed a trainable neuron model that can be expressed in a recurrent form. Finally, we trained its simulation with an artificially generated dataset of dog postures and implemented it as hardware that showed promising energetic properties. This paper is the full text of the research, presented at the 20th International Conference on Artificial Intelligence and Soft Computing Web System (ICAISC 2021)

cs.LG↗

arXiv2021

Linear Constraints Learning for Spiking Neurons

We introduce a new supervised learning algorithm based to train spiking neural networks for classification. The algorithm overcomes a limitation of existing multi-spike learning methods: it solves the problem of interference between interacting output spikes during a learning trial. This problem of learning interference causes learning performance in existing approaches to decrease as the number of output spikes increases, and represents an important limitation in existing multi-spike learning approaches. We address learning interference by introducing a novel mechanism to balance the magnitudes of weight adjustments during learning, which in theory allows every spike to simultaneously converge to their desired timings. Our results indicate that our method achieves significantly higher memory capacity and faster convergence compared to existing approaches for multi-spike classification. In the ubiquitous Iris and MNIST datasets, our algorithm achieves competitive predictive performance with state-of-the-art approaches.

cs.NE↗

arXiv2020

Constraints on Hebbian and STDP learned weights of a spiking neuron

We analyse mathematically the constraints on weights resulting from Hebbian and STDP learning rules applied to a spiking neuron with weight normalisation. In the case of pure Hebbian learning, we find that the normalised weights equal the promotion probabilities of weights up to correction terms that depend on the learning rate and are usually small. A similar relation can be derived for STDP algorithms, where the normalised weight values reflect a difference between the promotion and demotion probabilities of the weight. These relations are practically useful in that they allow checking for convergence of Hebbian and STDP algorithms. Another application is novelty detection. We demonstrate this using the MNIST dataset.

cs.LG↗

arXiv2020

A thermodynamically consistent chemical spiking neuron capable of autonomous Hebbian learning

We propose a fully autonomous, thermodynamically consistent set of chemical reactions that implements a spiking neuron. This chemical neuron is able to learn input patterns in a Hebbian fashion. The system is scalable to arbitrarily many input channels. We demonstrate its performance in learning frequency biases in the input as well as correlations between different input channels. Efficient computation of time-correlations requires a highly non-linear activation function. The resource requirements of a non-linear activation function are discussed. In addition to the thermodynamically consistent model of the CN, we also propose a biologically plausible version that could be engineered in a synthetic biology context.

cs.NE↗

arXiv2020

Minimal spiking neuron for solving multi-label classification tasks

The Multi-Spike Tempotron (MST) is a powerful single spiking neuron model that can solve complex supervised classification tasks. While powerful, it is also internally complex, computationally expensive to evaluate, and not suitable for neuromorphic hardware. Here we aim to understand whether it is possible to simplify the MST model, while retaining its ability to learn and to process information. To this end, we introduce a family of Generalised Neuron Models (GNM) which are a special case of the Spike Response Model and much simpler and cheaper to simulate than the MST. We find that over a wide range of parameters the GNM can learn at least as well as the MST. We identify the temporal autocorrelation of the membrane potential as the single most important ingredient of the GNM which enables it to classify multiple spatio-temporal patterns. We also interpret the GNM as a chemical system, thus conceptually bridging computation by neural networks with molecular information processing. We conclude the paper by proposing alternative training approaches for the GNM including error trace learning and error backpropagation.

cs.NE↗

arXiv2018

A thermodynamically consistent model of finite state machines

Finite state machines (FSMs) are a theoretically and practically important model of computation. We propose a general, thermodynamically consistent model of FSMs and characterise the resource requirements of these machines. We model FSMs as time-inhomogeneous Markov chains. The computation is driven by instantaneous manipulations of the energy levels of the states. We calculate the entropy production of the machine, its error probability, and the time required to complete one update step. We find that a sequence of generalised bit-setting operations is sufficient to implement any FSM.

cond-mat.stat-mech↗

arXiv2018

Information ratchets exploiting spatially structured information reservoirs

Information ratchets extract work from heat baths using low entropy information reservoirs. We find that the structure of the reservoirs affects both their performance and phase diagram. By deriving exact probabilities of recurrence in d-dimensional reservoirs we calculate the net extraction of work. Performance is characterised by two critical dimensions (i) For d < 3 driven exploration of the reservoir is essential to extract work (ii) For d > 4 purely diffusive exploration is optimal. This has important consequences for computation where sequential operations constitute a 1D path.

cond-mat.stat-mech↗

arXiv2018

The thermodynamics of quasi-deterministic digital computers

It is now well established that there is no lower bound for the energy dissipated during a computation. The relevance of the zero-energy limit is unclear, however, because it entails computations that are unreliable or infinitely slow, or both. In contrast the thermodynamic costs of deterministic computations that complete in finite time is less well understood. We propose a model of universal computation based on Markovian stochastic processes. While strictly deterministic computation is not possible in such systems, we show that the scaling of the energy consumption in relation to the accuracy of the computation is benign. This enables quasi-deterministic computation at modest cost in energy and completing within finite time.

cond-mat.stat-mech↗

arXiv2018

Performance limits and trade-offs in entropy-driven biochemical computers

The properties and fundamental limits of chemical computers have recently attracted significant interest as a model of computation, an unifying principle of cellular organisation and in the context of bio-engineering. As of yet, research in this topic is based on case-studies. There exists no generally accepted criterion to distinguish between chemical processes that compute and those that do not. Here, the concept of entropy driven computer (EDC) is proposed as a general model of chemical computation. It is found that entropy driven computation is subject to a trade-off between accuracy and entropy production, but unlike many biological systems, there are no trade-offs involving time. The latter only arise when it is taken into account that the observation of the state of the EDC is not energy neutral, but comes at a cost. The significance of this conclusion in relation to biological systems is discussed. Three examples of biological computers, including an implementation of a neural network as an EDC are given.

physics.bio-ph↗