Searcharxiv⌕ Search

arXiv subjects

Ella Gale

Publications and source records attributed to Ella Gale.

At least 19 recordsLinked to original sources
arXiv2024

Disconnection Rules are Complete for Chemical Reactions

We provide a category theoretical framework capturing two approaches to graph-based models of chemistry: formal reactions and disconnection rules. We model a translation from the latter to the former as a functor, which is faithful, and full up to isomorphism. This allows us to state, as our main result, that the disconnection rules are sound, complete and universal with respect to the reactions. Concretely, this means that every reaction can be decomposed into a sequence of disconnection rules in an essentially unique way. This provides a uniform way to store reaction data, and gives an algorithmic interface between (forward) reaction prediction and (backward) reaction search or retrosynthesis.

cs.LO↗
arXiv2023

A Categorical Model for Retrosynthetic Reaction Analysis

We introduce a mathematical framework for retrosynthetic analysis, an important research method in synthetic chemistry. Our approach represents molecules and their interaction using string diagrams in layered props - a recently introduced categorical model for partial explanations in scientific reasoning. Such principled approach allows one to model features currently not available in automated retrosynthesis tools, such as chirality, reaction environment and protection-deprotection steps.

cs.LO↗
arXiv2023

Cluster Flow: how a hierarchical clustering layer make allows deep-NNs more resilient to hacking, more human-like and easily implements relational reasoning

Despite the huge recent breakthroughs in neural networks (NNs) for artificial intelligence (specifically deep convolutional networks) such NNs do not achieve human-level performance: they can be hacked by images that would fool no human and lack `common sense'. It has been argued that a basis of human-level intelligence is mankind's ability to perform relational reasoning: the comparison of different objects, measuring similarity, grasping of relations between objects and the converse, figuring out the odd one out in a set of objects. Mankind can even do this with objects they have never seen before. Here we show how ClusterFlow, a semi-supervised hierarchical clustering framework can operate on trained NNs utilising the rich multi-dimensional class and feature data found at the pre-SoftMax layer to build a hyperspacial map of classes/features and this adds more human-like functionality to modern deep convolutional neural networks. We demonstrate this with 3 tasks. 1. the statistical learning based `mistakes' made by infants when attending to images of cats and dogs. 2. improving both the resilience to hacking images and the accurate measure of certainty in deep-NNs. 3. Relational reasoning over sets of images, including those not known to the NN nor seen before. We also demonstrate that ClusterFlow can work on non-NN data and deal with missing data by testing it on a Chemistry dataset. This work suggests that modern deep NNs can be made more human-like without re-training of the NNs. As it is known that some methods used in deep and convolutional NNs are not biologically plausible or perhaps even the best approach: the ClusterFlow framework can sit on top of any NN and will be a useful tool to add as NNs are improved in this regard.

cs.LG↗
arXiv2023

Shape is (almost) all!: Persistent homology features (PHFs) are an information rich input for efficient molecular machine learning

3-D shape is important to chemistry, but how important? Machine learning works best when the inputs are simple and match the problem well. Chemistry datasets tend to be very small compared to those generally used in machine learning so we need to get the most from each datapoint. Persistent homology measures the topological shape properties of point clouds at different scales and is used in topological data analysis. Here we investigate what persistent homology captures about molecular structure and create persistent homology features (PHFs) that encode a molecule's shape whilst losing most of the symbolic detail like atom labels, valence, charge, bonds etc. We demonstrate the usefulness of PHFs on a series of chemical datasets: QM7, lipophilicity, Delaney and Tox21. PHFs work as well as the best benchmarks. PHFs are very information dense and much smaller than other encoding methods yet found, meaning ML algorithms are much more energy efficient. PHFs success despite losing a large amount of chemical detail highlights how much of chemistry can be simplified to topological shape.

cs.LG↗
arXiv2023

Icospherical Chemical Objects (ICOs) allow for chemical data augmentation and maintain rotational, translation and permutation invariance

Dataset augmentation is a common way to deal with small datasets; Chemistry datasets are often small. Spherical convolutional neural networks (SphNNs) and Icosahedral neural networks (IcoNNs) are a type of geometric machine learning algorithm that maintains rotational symmetry. Molecular structure has rotational invariance and is inherently 3-D, and thus we need 3-D encoding methods to input molecular structure into machine learning. In this paper I present Icospherical Chemical Objects (ICOs) that enable the encoding of 3-D data in a rotationally invariant way which works with spherical or icosahedral neural networks and allows for dataset augmentation. I demonstrate the ICO featurisation method on the following tasks: predicting general molecular properties, predicting solubility of drug like molecules and the protein binding problem and find that ICO and SphNNs perform well on all problems.

cs.LG↗
arXiv2016

TiO$_2$-based Memristors and ReRAM: Materials, Mechanisms and Models (a Review)

The memristor is the fundamental non-linear circuit element, with uses in computing and computer memory. ReRAM (Resistive Random Access Memory) is a resistive switching memory proposed as a non-volatile memory. In this review we shall summarise the state of the art for these closely-related fields, concentrating on titanium dioxide, the well-utilised and archetypal material for both. We shall cover material properties, switching mechanisms and models to demonstrate what ReRAM and memristor scientists can learn from each other and examine the outlook for these technologies.

cond-mat.mtrl-sci↗
arXiv2016

Non-ideal memristors for a non-ideal world

Memristors have pinched hysteresis loops in the $V-I$ plane. Ideal memristors are everywhere non-linear, cross at zero and are rotationally symmetric. In this paper we extend memristor theory to produce different types of non-ideality and find that: including a background current (such as an ionic current) moves the crossing point away from zero; including a degradation resistance (that increases with experimental time) leads to an asymmetry; modelling a low resistance filament in parallel describes triangular $V-I$ curves with a straight-line low resistance state. A novel measurement of hysteresis asymmetry was introduced based on hysteresis and it was found that which lobe was bigger depended on the size of the breaking current relative to the memristance. The hysteresis varied differently with each type of non-ideality, suggesting that measurements of several device I-V curves and calculation of these parameters could give an indication of the underlying mechanism.

cond-mat.mes-hall↗
arXiv2015

Single Memristor Logic Gates: From NOT to a Full Adder

Memristors have been suggested as a novel route to neuromorphic computing based on the similarity between them and neurons (specifically synapses and ion pumps). The d.c. action of the memristor is a current spike which imparts a short-term memory to the device. Here it is demonstrated that this short-term memory works exactly like habituation (e.g. in \emph{Aplysia}). We elucidate the physical rules, based on energy conservation, governing the interaction of these current spikes: summation, `bounce-back', directionality and `diminishing returns'. Using these rules, we introduce 4 different logical systems to implement sequential logic in the memristor and demonstrate how sequential logic works by instantiating a NOT gate, an AND gate, an XOR gate and a Full Adder with a single memristor. The Full Adder makes use of the memristor's short-term memory to add together three binary values and outputs the sum, the carry digit and even the order they were input in. A memristor full adder also outputs the arithmetical sum of bits, allowing for a logically (but not physically) reversible system. Essentially, we can replace an input/output port with an extra time-step, allowing a single memristor to do a hither-to unexpectedly large amount of computation. This makes up for the memristor's slow operation speed and may relate to how neurons do a similarly-large computation with such slow operations speeds. We propose that using spiking logic, either in gates or as neuron-analogues, with plastic rewritable connections between them, would allow the building of a neuromorphic computer.

cs.ET↗
arXiv2015

Evolving Spiking Networks with Variable Resistive Memories

Neuromorphic computing is a brainlike information processing paradigm that requires adaptive learning mechanisms. A spiking neuro-evolutionary system is used for this purpose; plastic resistive memories are implemented as synapses in spiking neural networks. The evolutionary design process exploits parameter self-adaptation and allows the topology and synaptic weights to be evolved for each network in an autonomous manner. Variable resistive memories are the focus of this research; each synapse has its own conductance profile which modifies the plastic behaviour of the device and may be altered during evolution. These variable resistive networks are evaluated on a noisy robotic dynamic-reward scenario against two static resistive memories and a system containing standard connections only. Results indicate that the extra behavioural degrees of freedom available to the networks incorporating variable resistive memories enable them to outperform the comparative synapse types.

cs.NE↗
arXiv2014

Uniform and Piece-wise Uniform Fields in Memristor Models

The Strukov model was the phenomenological model that accompanied the announcement of the first recognised physical instantiation of the memristor and, as such, it has been widely used. This model described the motion of a boundary, $w$, between two types of inter-converting material, $R_{\mathrm{off}}$ and $R_{\mathrm{on}}$, seemingly under a uniform field across the entire device. In fact, what was intended was a field with a discontinuity at $w$, that was uniform between $0<x<w$. In this paper we show that the discontinuity is required for the Strukov model derivation to be completed, and thus the derivation as given does not describe a situation with a uniform field across the entire device. The discontinuity can be described as a Heaviside function, $H$, located on $w$, for which there are three common single-valued approximations for $H(w)$. The Strukov model as intended includes an approximation for the Heaviside function (the field is taken to be the same as that across the $R_{\mathrm{on}}$ part of the device). We compare approximations and give solutions. We then extend the description of the field to a more-realistic continuously varying sigmoidal transition between two uniform fields and demonstrate that the centro-symmetric approximation model (taking the field as being the average of the fields across $R_{\mathrm{on}}$ and $R_{\mathrm{off}}$) is a better single-point model of that situation: the other two approximations over or underestimate the field.

cs.ET↗
arXiv2014

Slime Mould Memristors

In laboratory experiments we demonstrate that protoplasmic tubes of acellular slime mould \emph{Physarum polycephalum} show current versus voltage profiles consistent with memristive systems and that the effect is due to the living protoplasm of the mould. This complements previous findings on memristive properties of other living systems (human skin and blood) and contributes to development of self-growing bio-electronic circuits. Distinctive asymmetric $V$-$I$ curves which were occasionally observed when the internal current is on the same order as the driven current, are well-modelled by the concept of active memristors.

cs.ET↗
arXiv2014

Emergent Spiking in Non-Ideal Memristor Networks

Memristors have uses as artificial synapses and perform well in this role in simulations with artificial spiking neurons. Our experiments show that memristor networks natively spike and can exhibit emergent oscillations and bursting spikes. Networks of near-ideal memristors exhibit behaviour similar to a single memristor and combine in circuits like resistors do. Spiking is more likely when filamentary memristors are used or the circuits have a higher degree of compositional complexity (i.e. a larger number of anti-series or anti-parallel interactions). 3-memristor circuits with the same memristor polarity (low compositional complexity) are stabilised and do not show spiking behaviour. 3-memristor circuits with anti-series and/or anti-parallel compositions show richer and more complex dynamics than 2-memristor spiking circuits. We show that the complexity of these dynamics can be quantified by calculating (using partial auto-correlation functions) the minimum order auto-regression function that could fit it. We propose that these oscillations and spikes may be similar phenomena to brainwaves and neural spike trains and suggest that these behaviours can be used to perform neuromorphic computation.

cond-mat.mtrl-sci↗
arXiv2014

The Memory-Conservation Theory of Memristance

The memristor, the recently discovered fundamental circuit element, is of great interest for neuromorphic computing, nonlinear electronics and computer memory. It is usually modelled either using Chua's equations, which lack material device properties, or using Strukov's phenomenological model (or models derived from it), which deviates from Chua's definitions due to the lack of a magnetic flux term. It is shown that by modelling the magnetostatics of the memory-holding ionic current (oxygen vacancies in the Strukov memristor), the memristor's magnetic flux can be identified as the flux arising from the ions. This leads to a novel theory of memristance consisting of two components: 1. A memory function which describes how the memristance, as felt by the ions, affects the conducting electrons located in the `on' part of the device; 2. A conservation function which describes the time-varying resistance in the `off' part of the device. This model allows for a straight-forward incorporation of the ions within the electronic theory and relates Chua's constitutive definition of a memristor with device material properties for the first time.

cond-mat.mtrl-sci↗
arXiv2014

Does the D.C. Response of Memristors Allow Robotic Short-Term Memory and a Possible Route to Artificial Time Perception?

Time perception is essential for task switching, and in the mammalian brain appears alongside other processes. Memristors are electronic components used as synapses and as models for neurons. The d.c. response of memristors can be considered as a type of short-term memory. Interactions of the memristor d.c. response within networks of memristors leads to the emergence of oscillatory dynamics and intermittent spike trains, which are similar to neural dynamics. Based on this data, the structure of a memristor network control for a robot as it undergoes task switching is discussed and it is suggested that these emergent network dynamics could improve the performance of role switching and learning in an artificial intelligence and perhaps create artificial time perception.

cs.RO↗
arXiv2014

The Short-term Memory (D.C. Response) of the Memristor Demonstrates the Causes of the Memristor Frequency Effect

A memristor is often identified by showing its distinctive pinched hysteresis curve and testing for the effect of frequency. The hysteresis size should relate to frequency and shrink to zero as the frequency approaches infinity. Although mathematically understood, the material causes for this are not well known. The d.c. response of the memristor is a decaying curve with its own timescale. We show via mathematical reasoning that this decaying curve when transformed to a.c. leads to the frequency effect by considering a descretized curve. We then demonstrate the validity of this approach with experimental data from two different types of memristors.

cs.ET↗
arXiv2014

Connecting Spiking Neurons to a Spiking Memristor Network Changes the Memristor Dynamics

Memristors have been suggested as neuromorphic computing elements. Spike-time dependent plasticity and the Hodgkin-Huxley model of the neuron have both been modelled effectively by memristor theory. The d.c. response of the memristor is a current spike. Based on these three facts we suggest that memristors are well-placed to interface directly with neurons. In this paper we show that connecting a spiking memristor network to spiking neuronal cells causes a change in the memristor network dynamics by: removing the memristor spikes, which we show is due to the effects of connection to aqueous medium; causing a change in current decay rate consistent with a change in memristor state; presenting more-linear $I-t$ dynamics; and increasing the memristor spiking rate, as a consequence of interaction with the spiking neurons. This demonstrates that neurons are capable of communicating directly with memristors, without the need for computer translation.

cs.ET↗
arXiv2014

Is Spiking Logic the Route to Memristor-Based Computers?

Memristors have been suggested as a novel route to neuromorphic computing based on the similarity between neurons (synapses and ion pumps) and memristors. The D.C. action of the memristor is a current spike, which we think will be fruitful for building memristor computers. In this paper, we introduce 4 different logical assignations to implement sequential logic in the memristor and introduce the physical rules, summation, `bounce-back', directionality and `diminishing returns', elucidated from our investigations. We then demonstrate how memristor sequential logic works by instantiating a NOT gate, an AND gate and a Full Adder with a single memristor. The Full Adder makes use of the memristor's memory to add three binary values together and outputs the value, the carry digit and even the order they were input in.

cs.ET↗