Luca Manneschi

Luca Manneschi

Auto-generated profile, based on publicly available information

Luca Manneschi

School of Computer Science, University of Sheffield, Sheffield, S10 2TN, United Kingdom

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
23 January 2023 in Applied Physics Letters
Verified
Journal Article
Open access
PDF file
DADan A. Allwood
Dan A. Allwood
MEMatthew O. A. Ellis
Matthew O. A. Ellis
DGDavid Griffin
David Griffin
THThomas J. Hayward
Thomas J. Hayward
LMLuca Manneschi
Luca Manneschi
CWChester Wringe
Chester Wringe
Neural networks have revolutionized the area of artificial intelligence and introduced transformative applications to almost every scientific field and industry. However, this success comes at a great price; the energy requirements for training advanced models are unsustainable. One promising way to address this pressing issue is by developing low-energy neuromorphic hardware that directly supports the algorithm's requirements. The intrinsic non-volatility, non-linearity, and memory of spintronic devices make them appealing candidates for neuromorphic devices. Here, we focus on the reservoir computing paradigm, a recurrent network with a simple training algorithm suitable for computation with spintronic devices since they can provide the properties of non-linearity and memory. We review technologies and methods for developing neuromorphic spintronic devices and conclude with critical open issues to address before such devices become widely used.

16 August 2021 in IEEE Transactions on Neural Networks and Learning Systems
Verified
Journal Article
Open access
LMLuca Manneschi
Luca Manneschi
ALAndrew C. Lin
Andrew C. Lin
EVEleni Vasilaki
Eleni Vasilaki
``Sparse'' neural networks, in which relatively few neurons or connections are active, are common in both machine learning and neuroscience. While, in machine learning, ``sparsity'' is related to a penalty term that leads to some connecting weights becoming small or zero, in biological brains, sparsity is often created when high spiking thresholds prevent neuronal activity. Here, we introduce sparsity into a reservoir computing network via neuron-specific learnable thresholds of activity, allowing neurons with low thresholds to contribute to decision-making but suppressing information from neurons with high thresholds. This approach, which we term ``SpaRCe,'' optimizes the sparsity level of the reservoir without affecting the reservoir dynamics. The read-out weights and the thresholds are learned by an online gradient rule that minimizes an error function on the outputs of the network. Threshold learning occurs by the balance of two opposing forces: reducing interneuronal correlations in the reservoir by deactivating redundant neurons, while increasing the activity of neurons participating in correct decisions. We test SpaRCe on classification problems and find that threshold learning improves performance compared to standard reservoir computing. SpaRCe alleviates the problem of catastrophic forgetting, a problem most evident in standard echo state networks (ESNs) and recurrent neural networks in general, due to increasing the number of task-specialized neurons that are included in the network decisions.

11 February 2022 in IEEE Robotics and Automation Letters
Verified
Journal Article
Closed access
AÖAnl Özdemir
Anl Özdemir
MSMark Scerri
Mark Scerri
ABAndrew B. Barron
Andrew B. Barron
APAndrew Philippides
Andrew Philippides
MMMichael Mangan
Michael Mangan
LMLuca Manneschi
Luca Manneschi
Recognising previously visited locations is an important, but unsolved, task in autonomous navigation. Current visual place recognition (VPR) benchmarks typically challenge models to recover the position of a query image (or images) from sequential datasets that include both spatial and temporal components. Recently, Echo State Network (ESN) varieties have proven particularly powerful at solving machine learning tasks that require spatio-temporal modelling. These networks are simple, yet powerful neural architectures that—exhibiting memory over multiple time-scales and non-linear high-dimensional representations—can discover temporal relations in the data while still maintaining linearity in the learning time. In this letter, we present a series of ESNs and analyse their applicability to the VPR problem. We report that the addition of ESNs to pre-processed convolutional neural networks led to a dramatic boost in performance in comparison to non-recurrent networks in five out of six standard benchmarks (GardensPoint, SPEDTest, ESSEX3IN1, Oxford RobotCar, and Nordland), demonstrating that ESNs are able to capture the temporal structure inherent in VPR problems. Moreover, we show that models that include ESNs can outperform class-leading VPR models which also exploit the sequential dynamics of the data. Finally, our results demonstrate that ESNs improve generalisation abilities, robustness, and accuracy further supporting their suitability to VPR applications.

16 October 2025 in Nature Communications
Verified
Journal Article
Open access
PDF file
LMLuca Manneschi
Luca Manneschi
IVIan T. Vidamour
Ian T. Vidamour
KSKilian D. Stenning
Kilian D. Stenning
CSCharles Swindells
Charles Swindells
GVGuru Venkat
Guru Venkat
EVEleni Vasilaki
Eleni Vasilaki
In materio computing offers the potential for widespread embodied intelligence by leveraging the intrinsic dynamics of complex systems for efficient sensing, processing, and interaction. While individual devices offer basic data processing capabilities, networks of interconnected devices can perform more complex and varied tasks. However, designing such networks for dynamic tasks is challenging in the absence of physical models and accurate characterization of device noise. We introduce the Noise-Aware Dynamic Optimization (NADO) framework for training networks of dynamical devices, using Neural Stochastic Differential Equations (Neural-SDEs) as differentiable digital twins to capture both the dynamics and stochasticity of devices with intrinsic memory. Our approach combines backpropagation through time with cascade learning, enabling effective exploitation of the temporal properties of physical devices. We validate this method on networks of spintronic devices across both temporal classification and regression tasks. By decoupling device model training from network connectivity optimization, our framework reduces data requirements and enables robust, gradient-based programming of dynamical devices without requiring analytical descriptions of their behaviour.

5 August 2022 in PLOS Computational Biology
Verified
Journal Article
Open access
PDF file
LMLuca Manneschi
Luca Manneschi
GGGuido Gigante
Guido Gigante
EVEleni Vasilaki
Eleni Vasilaki
PGPaolo Del Giudice
Paolo Del Giudice
We postulate that three fundamental elements underlie a decision making process: perception of time passing, information processing in multiple timescales and reward maximisation. We build a simple reinforcement learning agent upon these principles that we train on a random dot-like task. Our results, similar to the experimental data, demonstrate three emerging signatures. (1) signal neutrality: insensitivity to the signal coherence in the interval preceding the decision. (2) Scalar property: the mean of the response times varies widely for different signal coherences, yet the shape of the distributions stays almost unchanged. (3) Collapsing boundaries: the “effective” decision-making boundary changes over time in a manner reminiscent of the theoretical optimal. Removing the perception of time or the multiple timescales from the model does not preserve the distinguishing signatures. Our results suggest an alternative explanation for signal neutrality. We propose that it is not part of motor planning. It is part of the decision-making process and emerges from information processing on multiple timescales.

2 September 2021 in bioRxiv
Verified
Preprint
Open access
PDF file
LMLuca Manneschi
Luca Manneschi
GGGuido Gicante
Guido Gicante
PGPaolo Del Giudice
Paolo Del Giudice
EVEleni Vasilaki
Eleni Vasilaki
Experiments and models in perceptual decision-making point to a key role of an integration process that accumulates sensory evidence over time. We endow a probabilistic agent comprising several such integrators with widely spread time scales and let it learn, by trial-and-error, to weight the different filtered versions of a noisy signal. The agent discovers a strategy markedly different from the literature “standard”, according to which a decision made when the accumulated evidence hits a predetermined threshold. The agent instead decides during fleeting windows corresponding to the alignment of many integrators, akin to a majority vote. This strategy presents three distinguishing signatures. 1) Signal neutrality: a marked insensitivity to the signal coherence in the interval preceding the decision, as also observed in experiments. 2) Scalar property: the mean of the response times varies glaringly for different signal coherences, yet the shape of the distribution stays largely unchanged. 3) Collapsing boundaries: the agent learns to behave as if subject to a non-monotonic urgency signal, reminiscent in shape of the theoretically optimal. These three characteristics, which emerge from the interaction of a multi-scale learning agent with a highly volatile environment, are hallmarks, we argue, of an optimal decision strategy in challenging situations. As such, the present results may shed light on general information-processing principles leveraged by the brain itself. Author summary The rate of integration of sensory information prior to a decision-making process needs to be versatile and adaptable to different situations. While driving can require quick reactions, evaluating the authenticity of a painting can require long observations, and consequently the concept of representations over multiple timescales appears necessary from an intuitive perspective. Nevertheless, there is a lack of theoretical research that exploits multiple timescales, despite the presence of a variety of integration rates have been experimentally observed. In the following work, we developed a decision-making model based on integrators with multiple characteristic times and analysed its behaviour on a highly volatile and biologically relevant task. Through trial and error and reward maximisation, the model discovers an effective strategy that is surprisingly different and more robust in comparison to the more “classical”, single time-scale approach. More importantly, the strategy learnt exhibits remarkable agreement with experimental findings, suggesting a fundamental role of multiple timescales for decision-making. Our model, despite being abstract, achieves a good degree of biological realism and perform robustly in different environments.