arXiv subjects

Stephen Clark

Publications and source records attributed to Stephen Clark.

At least 37 records · Page 2Linked to original sources

arXiv2020

Probing Emergent Semantics in Predictive Agents via Question Answering

Recent work has shown how predictive modeling can endow agents with rich knowledge of their surroundings, improving their ability to act in complex environments. We propose question-answering as a general paradigm to decode and understand the representations that such agents develop, applying our method to two recent approaches to predictive modeling -action-conditional CPC (Guo et al., 2018) and SimCore (Gregor et al., 2019). After training agents with these predictive objectives in a visually-rich, 3D environment with an assortment of objects, colors, shapes, and spatial configurations, we probe their internal state representations with synthetic (English) questions, without backpropagating gradients from the question-answering decoder into the agent. The performance of different agents when probed this way reveals that they learn to encode factual, and seemingly compositional, information about objects, properties and spatial relations from their physical environment. Our approach is intuitive, i.e. humans can easily interpret responses of the model as opposed to inspecting continuous vectors, and model-agnostic, i.e. applicable to any modeling approach. By revealing the implicit knowledge of objects, quantities, properties and relations acquired by agents as they learn, question-conditional agent probing can stimulate the design and development of stronger predictive learning objectives.

cs.AI↗

arXiv2020

Environmental drivers of systematicity and generalization in a situated agent

The question of whether deep neural networks are good at generalising beyond their immediate training experience is of critical importance for learning-based approaches to AI. Here, we consider tests of out-of-sample generalisation that require an agent to respond to never-seen-before instructions by manipulating and positioning objects in a 3D Unity simulated room. We first describe a comparatively generic agent architecture that exhibits strong performance on these tests. We then identify three aspects of the training regime and environment that make a significant difference to its performance: (a) the number of object/word experiences in the training set; (b) the visual invariances afforded by the agent's perspective, or frame of reference; and (c) the variety of visual input inherent in the perceptual aspect of the agent's perception. Our findings indicate that the degree of generalisation that networks exhibit can depend critically on particulars of the environment in which a given task is instantiated. They further suggest that the propensity for neural networks to generalise in systematic ways may increase if, like human children, those networks have access to many frames of richly varying, multi-modal observations as they learn.

cs.AI↗

arXiv2020

Who voted for a No Deal Brexit? A Composition Model of Great Britains 2019 European Parliamentary Elections

The purpose of this paper is to use the votes cast at the 2019 European elections held in United Kingdom to re-visit the analysis conducted subsequent to its 2016 European Union referendum vote. This exercise provides a staging post on public opinion as the United Kingdom moves to leave the European Union during 2020. A composition data analysis in a seemingly unrelated regression framework is adopted that respects the compositional nature of the vote outcome; each outcome is a share that adds up to 100% and each outcome is related to the alternatives. Contemporary explanatory data for each counting area is sourced from the themes of socio-demographics, employment, life satisfaction and place. The study find that there are still strong and stark divisions in the United Kingdom, defined by age, qualifications, employment and place. The use of a compositional analysis approach produces challenges in regards to the interpretation of these models, but marginal plots are seen to aid the interpretation somewhat.

physics.soc-ph↗

arXiv2019

Understanding Early Word Learning in Situated Artificial Agents

Neural network-based systems can now learn to locate the referents of words and phrases in images, answer questions about visual scenes, and execute symbolic instructions as first-person actors in partially-observable worlds. To achieve this so-called grounded language learning, models must overcome challenges that infants face when learning their first words. While it is notable that models with no meaningful prior knowledge overcome these obstacles, researchers currently lack a clear understanding of how they do so, a problem that we attempt to address in this paper. For maximum control and generality, we focus on a simple neural network-based language learning agent, trained via policy-gradient methods, which can interpret single-word instructions in a simulated 3D world. Whilst the goal is not to explicitly model infant word learning, we take inspiration from experimental paradigms in developmental psychology and apply some of these to the artificial agent, exploring the conditions under which established human biases and learning effects emerge. We further propose a novel method for visualising semantic representations in the agent.

cs.CL↗

arXiv2019

Neural Generative Rhetorical Structure Parsing

Rhetorical structure trees have been shown to be useful for several document-level tasks including summarization and document classification. Previous approaches to RST parsing have used discriminative models; however, these are less sample efficient than generative models, and RST parsing datasets are typically small. In this paper, we present the first generative model for RST parsing. Our model is a document-level RNN grammar (RNNG) with a bottom-up traversal order. We show that, for our parser's traversal order, previous beam search algorithms for RNNGs have a left-branching bias which is ill-suited for RST parsing. We develop a novel beam search algorithm that keeps track of both structure- and word-generating actions without exhibiting this branching bias and results in absolute improvements of 6.8 and 2.9 on unlabelled and labelled F1 over previous algorithms. Overall, our generative model outperforms a discriminative model with the same features by 2.6 F1 points and achieves performance comparable to the state-of-the-art, outperforming all published parsers from a recent replication study that do not use additional training data.

cs.CL↗

arXiv2019

Scalable Syntax-Aware Language Models Using Knowledge Distillation

Prior work has shown that, on small amounts of training data, syntactic neural language models learn structurally sensitive generalisations more successfully than sequential language models. However, their computational complexity renders scaling difficult, and it remains an open question whether structural biases are still necessary when sequential models have access to ever larger amounts of training data. To answer this question, we introduce an efficient knowledge distillation (KD) technique that transfers knowledge from a syntactic language model trained on a small corpus to an LSTM language model, hence enabling the LSTM to develop a more structurally sensitive representation of the larger training data it learns from. On targeted syntactic evaluations, we find that, while sequential LSTMs perform much better than previously reported, our proposed technique substantially improves on this baseline, yielding a new state of the art. Our findings and analysis affirm the importance of structural biases, even in models that learn from large amounts of data.

cs.CL↗

arXiv2019

Factorising AMR generation through syntax

Generating from Abstract Meaning Representation (AMR) is an underspecified problem, as many syntactic decisions are not constrained by the semantic graph. To explicitly account for this underspecification, we break down generating from AMR into two steps: first generate a syntactic structure, and then generate the surface form. We show that decomposing the generation process this way leads to state-of-the-art single model performance generating from AMR without additional unlabelled data. We also demonstrate that we can generate meaning-preserving syntactic paraphrases of the same AMR graph, as judged by humans.

cs.CL↗

arXiv2018

Latent Tree Learning with Differentiable Parsers: Shift-Reduce Parsing and Chart Parsing

Latent tree learning models represent sentences by composing their words according to an induced parse tree, all based on a downstream task. These models often outperform baselines which use (externally provided) syntax trees to drive the composition order. This work contributes (a) a new latent tree learning model based on shift-reduce parsing, with competitive downstream performance and non-trivial induced trees, and (b) an analysis of the trees learned by our shift-reduce model and by a chart-based model.

cs.CL↗

arXiv2018

Emergent Communication through Negotiation

Multi-agent reinforcement learning offers a way to study how communication could emerge in communities of agents needing to solve specific problems. In this paper, we study the emergence of communication in the negotiation environment, a semi-cooperative model of agent interaction. We introduce two communication protocols -- one grounded in the semantics of the game, and one which is \textit{a priori} ungrounded and is a form of cheap talk. We show that self-interested agents can use the pre-grounded communication channel to negotiate fairly, but are unable to effectively use the ungrounded channel. However, prosocial agents do learn to use cheap talk to find an optimal negotiating strategy, suggesting that cooperation is necessary for language to emerge. We also study communication behaviour in a setting where one agent interacts with agents in a community with different levels of prosociality and show how agent identifiability can aid negotiation.

cs.AI↗

arXiv2018

Emergence of Linguistic Communication from Referential Games with Symbolic and Pixel Input

The ability of algorithms to evolve or learn (compositional) communication protocols has traditionally been studied in the language evolution literature through the use of emergent communication tasks. Here we scale up this research by using contemporary deep learning methods and by training reinforcement-learning neural network agents on referential communication games. We extend previous work, in which agents were trained in symbolic environments, by developing agents which are able to learn from raw pixel data, a more challenging and realistic input representation. We find that the degree of structure found in the input data affects the nature of the emerged protocols, and thereby corroborate the hypothesis that structured compositional language is most likely to emerge when agents perceive the world as being structured.

cs.AI↗

arXiv2017

Jointly Learning Sentence Embeddings and Syntax with Unsupervised Tree-LSTMs

We introduce a neural network that represents sentences by composing their words according to induced binary parse trees. We use Tree-LSTM as our composition function, applied along a tree structure found by a fully differentiable natural language chart parser. Our model simultaneously optimises both the composition function and the parser, thus eliminating the need for externally-provided parse trees which are normally required for Tree-LSTM. It can therefore be seen as a tree-based RNN that is unsupervised with respect to the parse trees. As it is fully differentiable, our model is easily trained with an off-the-shelf gradient descent method and backpropagation. We demonstrate that it achieves better performance compared to various supervised Tree-LSTM architectures on a textual entailment task and a reverse dictionary task.

cs.CL↗

arXiv2017

Latent Variable Dialogue Models and their Diversity

We present a dialogue generation model that directly captures the variability in possible responses to a given input, which reduces the `boring output' issue of deterministic dialogue models. Experiments show that our model generates more diverse outputs than baseline models, and also generates more consistently acceptable output than sampling from a deterministic encoder-decoder model.

cs.CL↗

arXiv2016

Virtual Embodiment: A Scalable Long-Term Strategy for Artificial Intelligence Research

Meaning has been called the "holy grail" of a variety of scientific disciplines, ranging from linguistics to philosophy, psychology and the neurosciences. The field of Artifical Intelligence (AI) is very much a part of that list: the development of sophisticated natural language semantics is a sine qua non for achieving a level of intelligence comparable to humans. Embodiment theories in cognitive science hold that human semantic representation depends on sensori-motor experience; the abundant evidence that human meaning representation is grounded in the perception of physical reality leads to the conclusion that meaning must depend on a fusion of multiple (perceptual) modalities. Despite this, AI research in general, and its subdisciplines such as computational linguistics and computer vision in particular, have focused primarily on tasks that involve a single modality. Here, we propose virtual embodiment as an alternative, long-term strategy for AI research that is multi-modal in nature and that allows for the kind of scalability required to develop the field coherently and incrementally, in an ethically responsible fashion.

cs.AI↗

arXiv2016

On discrete symplectic systems: Associated maximal and minimal linear relations and nonhomogeneous problems

In this paper we characterize the definiteness of the discrete symplectic system, study a nonhomogeneous discrete symplectic system, and introduce the minimal and maximal linear relations associated with these systems. Fundamental properties of the corresponding deficiency indices, including a relationship between the number of square summable solutions and the dimension of the defect subspace, are also derived. Moreover, a sufficient condition for the existence of a densely defined operator associated with the symplectic system is provided.

math.SP↗

arXiv2016

Characterization of self-adjoint extensions for discrete symplectic systems

All self-adjoint extensions of minimal linear relation associated with the discrete symplectic system are characterized. Especially, for the scalar case on a finite discrete interval some equivalent forms and the uniqueness of the given expression are discussed and the Krein--von Neumann extension is described explicitly. In addition, a limit point criterion for symplectic systems is established. The result partially generalizes even a classical limit point criterion for the second order Sturm--Liouville difference equations.

math.SP↗

arXiv2015

Principal Solutions Revisited

The main objective of this paper is to identify principal solutions associated with Sturm-Liouville operators on arbitrary open intervals $(a,b) \subseteq \mathbb{R}$, as introduced by Leighton and Morse in the scalar context in 1936 and by Hartman in the matrix-valued situation in 1957, with Weyl-Titchmarsh solutions, as long as the underlying Sturm-Liouville differential expression is nonoscillatory (resp., disconjugate or bounded from below near an endpoint) and in the limit point case at the endpoint in question. In addition, we derive an explicit formula for Weyl-Titchmarsh functions in this case (the latter appears to be new in the matrix-valued context).

math.SP↗

arXiv2014

Using Sentence Plausibility to Learn the Semantics of Transitive Verbs

The functional approach to compositional distributional semantics considers transitive verbs to be linear maps that transform the distributional vectors representing nouns into a vector representing a sentence. We conduct an initial investigation that uses a matrix consisting of the parameters of a logistic regression classifier trained on a plausibility task as a transitive verb function. We compare our method to a commonly used corpus-based method for constructing a verb matrix and find that the plausibility training may be more effective for disambiguation tasks.

cs.CL↗

arXiv2014

The Frobenius anatomy of word meanings II: possessive relative pronouns

Within the categorical compositional distributional model of meaning, we provide semantic interpretations for the subject and object roles of the possessive relative pronoun `whose'. This is done in terms of Frobenius algebras over compact closed categories. These algebras and their diagrammatic language expose how meanings of words in relative clauses interact with each other. We show how our interpretation is related to Montague-style semantics and provide a truth-theoretic interpretation. We also show how vector spaces provide a concrete interpretation and provide preliminary corpus-based experimental evidence. In a prequel to this paper, we used similar methods and dealt with the case of subject and object relative pronouns.

cs.CL↗