In light of recent advances in large language models (LLMs), the expectations for the next generation of virtual assistants include enhanced naturalness and adaptability across diverse usage scenarios. However, the creation of high-quality annotated data for Task-Oriented Dialog (TOD) is recognized to be slow and costly. To address these challenges, we introduce Task-Oriented Automatic Dialogs (TOAD), a novel and scalable TOD dataset along with its automatic generation pipeline. The TOAD dataset simulates realistic app context interaction and provide a variety of system response style options. Two aspects of system response styles are considered, verbosity level and users' expression mirroring. We benchmark TOAD on two response generation tasks, and the results show that modeling more verbose responses or responses without user expression mirroring is more challenging.(1)

Dialog modelling faces a difficult trade-off. Models are trained on a large amount of text, yet their responses need to be limited to a desired scope and style of a dialog agent. Because the datasets used to achieve the former contain language that is not compatible with the latter, pre-trained dialog models are fine-tuned on smaller curated datasets. However, the fine-tuning process robs them of the ability to produce diverse responses, eventually reducing them to dull conversation partners. In this paper we investigate if prompting can mitigate the above trade-off. Specifically, we experiment with conditioning the prompt on the query, rather than training a single prompt for all queries. By following the intuition that freezing the pre-trained language model will conserve its expressivity, we find that compared to fine-tuning, prompting can achieve a higher BLEU score and substantially improve the diversity and novelty of the responses.

Adversarial attacks aim to change the predic-001 tions of deep neural network models, while re-002 maining unnoticed by the user. In this study, we 003 investigate the robustness of visually grounded 004 dialog models towards textual attacks. First, 005 to understand how different input components 006 can mitigate the attack. Our results show that 007 dialog history is important for model robust-008 ness: models encoding history are more robust, 009 and when launching an attack on history, model 010 prediction becomes more uncertain. This is in 011 contrast to prior work which finds that dialog 012 history is negligible for model performance. 013 We also evaluate how to generate adversarial 014 examples which successfully attack the model 015 but remain undetected by the user. We find 016 that the textual, as well as the visual context 017 is important to generate attacks which appear 018 semantically coherent to humans. 019

Multi-lingual language models (LM), such as mBERT, XLM-R, mT5, mBART, have been remarkably successful in enabling natural language tasks in low-resource languages through cross-lingual transfer from high-resource ones. In this work, we try to better understand how such models, specifically mT5, transfer *any* linguistic and semantic knowledge across languages, even though no explicit cross-lingual signals are provided during pre-training. Rather, only unannotated texts from each language are presented to the model separately and independently of one another, and the model appears to implicitly learn cross-lingual connections. This raises several questions that motivate our study, such as: Are the cross-lingual connections between every language pair equally strong? What properties of source and target language impact the strength of cross-lingual transfer? Can we quantify the impact of those properties on the cross-lingual transfer? In our investigation, we analyze a pre-trained mT5 to discover the attributes of cross-lingual connections learned by the model. Through a statistical interpretation framework over 90 language pairs across three tasks, we show that transfer performance can be modeled by a few linguistic and data-derived features. These observations enable us to interpret cross-lingual understanding of the mT5 model. Through these observations, one can favorably choose the best source language for a task, and can anticipate its training data demands. A key finding of this work is that similarity of syntax, morphology and phonology are good predictors of cross-lingual transfer, significantly more than just the lexical similarity of languages. For a given language, we are able to predict zero-shot performance, that increases on a logarithmic scale with the number of few-shot target language data points.

Non-autoregressive generation (NAG) has recently attracted great attention due to its fast inference speed. However, the generation quality of existing NAG models still lags behind their autoregressive counterparts. In this work, we show that BERT can be employed as the backbone of a NAG model to greatly improve performance. Additionally, we devise mechanisms to alleviate the two common problems of vanilla NAG models: the inflexibility of prefixed output length and the conditional independence of individual token predictions. Lastly, to further increase the speed advantage of the proposed model, we propose a new decoding strategy, ratio-first, for applications where the output lengths can be approximately estimated beforehand. For a comprehensive evaluation, we test the proposed model on three text generation tasks, including text summarization, sentence compression and machine translation. Experimental results show that our model significantly outperforms existing non-autoregressive baselines and achieves competitive performance with many strong autoregressive models. In addition, we also conduct extensive analysis experiments to reveal the effect of each proposed component.

Recent developments in neural networks have led to the advance in data-to-text generation. However, the lack of ability of neural models to control the structure of generated output can be limiting in certain real-world applications. In this study, we propose a novel Plan-then-Generate (PlanGen) framework to improve the controllability of neural data-to-text models. Extensive experiments and analyses are conducted on two benchmark datasets, ToTTo and WebNLG. The results show that our model is able to control both the intra-sentence and inter-sentence structure of the generated output. Furthermore, empirical comparisons against previous state-of-the-art methods show that our model improves the generation quality as well as the output diversity as judged by human and automatic evaluations.

Paraphrase generation is an important and challenging NLG problem.In this work, we propose a new Identification-then-Aggregation (IA) framework to tackle this task.In the identification step, the input tokens are sorted into two groups by a novel Primary/Secondary Identification (PSI) algorithm.In the aggregation step, these groups are separately encoded, before being aggregated by a custom designed decoder, which autoregressively generates the paraphrased sentence.In extensive experiments on two benchmark datasets, we demonstrate that our model outperforms previous studies by a notable margin.We also show that the proposed approach can generate paraphrases in an interpretable and controllable way.

Natural language understanding (NLU) and natural language generation (NLG) are two fundamental and related tasks in building task-oriented dialogue systems with opposite objectives: NLU tackles the transformation from natural language to formal representations, whereas NLG does the reverse. A key to success in either task is parallel training data which is expensive to obtain at a large scale. In this work, we propose a generative model which couples NLU and NLG through a shared latent variable. This approach allows us to explore both spaces of natural language and formal representations, and facilitates information sharing through the latent space to eventually benefit NLU and NLG. Our model achieves state-of-the-art performance on two dialogue datasets with both flat and tree-structured formal representations. We also show that the model can be trained in a semi-supervised fashion by utilising unlabelled data to boost its performance.

Spoken dialogue systems allow humans to interact with machines using natural speech. As such, they have many benefits. By using speech as the primary communication medium, a computer interface can facilitate swift, human-like acquisition of information. In recent years, speech interfaces have become ever more popular, as is evident from the rise of personal assistants such as Siri, Google Now, Cortana and Amazon Alexa. Recently, data-driven machine learning methods have been applied to dialogue modelling and the results achieved for limited-domain applications are comparable to or out-perform traditional approaches. Methods based on Gaussian processes are particularly effective as they enable good models to be estimated from limited training data. Furthermore, they provide an explicit estimate of the uncertainty which is particularly useful for reinforcement learning. This article explores the additional steps that are necessary to extend these methods to model multiple dialogue domains. We show that Gaussian process reinforcement learning is an elegant framework that naturally supports a range of methods, including prior knowledge, Bayesian committee machines and multi-agent learning, for facilitating extensible and adaptable dialogue systems.

Teaching machines to accomplish tasks by conversing naturally with humans is challenging. Currently, developing task-oriented dialogue systems requires creating multiple components and typically this involves either a large amount of handcrafting, or acquiring costly labelled datasets to solve a statistical learning problem for each component. In this work we introduce a neural network-based text-in, text-out end-to-end trainable goal-oriented dialogue system along with a new way of collecting dialogue data based on a novel pipe-lined Wizard-of-Oz framework. This approach allows us to develop dialogue systems easily and without making too many assumptions about the task at hand. The results show that the model can converse with human subjects naturally whilst helping them to accomplish tasks in a restaurant search domain.

Statistical Spoken Dialogue Systems have been around for many years. However, access to these systems has always been difficult as there is still no publicly available end-to-end system implementation. To alleviate this, we present PyDial, an open-source end-to-end statistical spoken dialogue system toolkit which provides implementations of statistical approaches for all dialogue system modules. Moreover, it has been extended to provide multidomain conversational functionality. It offers easy configuration, easy extensibility, and domain-independent implementations of the respective dialogue system modules. The toolkit is available for download under the Apache 2.0 license.

Existing speaking tests only require non-native speakers to engage in dialogue when the assessment is done by humans. This paper examines the viability of us-ing off-the-shelf systems for spoken dialogue and for speech grading to automate the holistic scoring of the conversational speech of non-native speakers of English

Moving from limited-domain natural language generation (NLG) to open domain is difficult because the number of semantic input combinations grows exponentially with the number of domains. Therefore, it is important to leverage existing resources and exploit similarities between domains to facilitate domain adaptation. In this paper, we propose a procedure to train multi-domain, Recurrent Neural Network-based (RNN) language generators via multiple adaptation steps. In this procedure, a model is first trained on counterfeited data synthesised from an out-of-domain dataset, and then fine tuned on a small set of in-domain utterances with a discriminative objective function. Corpus-based evaluation results show that the proposed procedure can achieve competitive performance in terms of BLEU score and slot error rate while significantly reducing the data needed to train generators in new, unseen domains. In subjective testing, human judges confirm that the procedure greatly improves generator performance when only a small amount of data is available in the domain.

In this work, we present a novel counter-fitting method which injects antonymy and synonymy constraints into vector space representations in order to improve the vectors' capability for judging semantic similarity. Applying this method to publicly available pre-trained word vectors leads to a new state of the art performance on the SimLex-999 dataset. We also show how the method can be used to tailor the word vector space for the downstream task of dialogue state tracking, resulting in robust improvements across different dialogue domains.

Recently a variety of LSTM-based conditional language models (LM) have been applied across a range of language generation tasks. In this work we study various model architectures and different ways to represent and aggregate the source information in an end-to-end neural dialogue system framework. A method called snapshot learning is also proposed to facilitate learning from supervised sequential signals by applying a companion cross-entropy objective function to the conditioning vector. The experimental and analytical results demonstrate firstly that competition occurs between the conditioning vector and the LM, and the differing architectures provide different trade-offs between the two. Secondly, the discriminative power and transparency of the conditioning vector is key to providing both model interpretability and better performance. Thirdly, snapshot learning leads to consistent performance improvements independent of which architecture is used.

We describe a two-step approach for dialogue management in task-oriented spoken dialogue systems. A unified neural network framework is proposed to enable the system to first learn by supervision from a set of dialogue data and then continuously improve its behaviour via reinforcement learning, all using gradient-based algorithms on one single model. The experiments demonstrate the supervised model's effectiveness in the corpus-based evaluation, with user simulation, and with paid human subjects. The use of reinforcement learning further improves the model's performance in both interactive settings, especially under higher-noise conditions.

The ability to compute an accurate reward function is essential for optimising a dialogue policy via reinforcement learning. In real-world applications, using explicit user feedback as the reward signal is often unreliable and costly to collect. This problem can be mitigated if the user's intent is known in advance or data is available to pre-train a task success predictor off-line. In practice neither of these apply for most real world applications. Here we propose an on-line learning framework whereby the dialogue policy is jointly trained alongside the reward model via active learning with a Gaussian process model. This Gaussian process operates on a continuous space dialogue representation generated in an unsupervised fashion using a recurrent neural network encoder-decoder. The experimental results demonstrate that the proposed framework is able to significantly reduce data annotation costs and mitigate noisy user feedback in dialogue policy learning.

Dialog state tracking is a key component of many modern dialog systems, most of which are designed with a single, well-defined domain in mind. This paper shows that dialog data drawn from different dialog domains can be used to train a general belief tracking model which can operate across all of these domains, exhibiting superior performance to each of the domain-specific models. We propose a training procedure which uses out-of-domain data to initialise belief tracking models for entirely new domains. This procedure leads to improvements in belief tracking performance regardless of the amount of in-domain data available for training the model.

The natural language generation (NLG) component of a spoken dialogue system (SDS) usually needs a substantial amount of handcrafting or a well-labeled dataset to be trained on.These limitations add significantly to development costs and make cross-domain, multi-lingual dialogue systems intractable.Moreover, human languages are context-aware.The most natural response should be directly learned from data rather than depending on predefined syntaxes or rules.This paper presents a statistical language generator based on a joint recurrent and convolutional neural network structure which can be trained on dialogue act-utterance pairs without any semantic alignments or predefined grammar trees.Objective metrics suggest that this new model outperforms previous methods under the same experimental conditions.Results of an evaluation by human judges indicate that it produces not only high quality but linguistically varied utterances which are preferred compared to n-gram and rule-based systems.