Papers by Raniero Romagnoli
arXiv (Cornell University), Sep 27, 2021
Incorporating explicit domain knowledge into neural-based task-oriented dialogue systems is an ef... more Incorporating explicit domain knowledge into neural-based task-oriented dialogue systems is an effective way to reduce the need of large sets of annotated dialogues. In this paper, we investigate how the use of explicit domain knowledge of conversational designers affects the performance of neural-based dialogue systems. To support this investigation, we propose the Conversational-Logic-Injection-in-Neural-Network system (CLINN) where explicit knowledge is coded in semilogical rules. By using CLINN, we evaluated semi-logical rules produced by a team of differently-skilled conversational designers. We experimented with the Restaurant topic of the MultiWOZ dataset. Results show that external knowledge is extremely important for reducing the need of annotated examples for conversational systems. In fact, rules from conversational designers used in CLINN significantly outperform a state-of-the-art neural-based dialogue system.
Proceedings of the Fourth Italian Conference on Computational Linguistics CLiC-it 2017
English. In this paper, we describe our neural network models for a commercial application on sen... more English. In this paper, we describe our neural network models for a commercial application on sentiment analysis. Different from academic work, which is oriented towards complex networks for achieving a marginal improvement, real scenarios require flexible and efficient neural models. The possibility to use the same models on different domains and languages plays an important role in the selection of the most appropriate architecture. We found that a small modification of the state-of-theart network according to academic benchmarks led to a flexible neural model that also preserves high accuracy.
Sequence-to-sequence neural networks are redesigning dialog managers for Conversational AI in ind... more Sequence-to-sequence neural networks are redesigning dialog managers for Conversational AI in industries. However, industrial applications impose two important constraints: training data are often scarce and the behavior of dialog managers should be strictly controlled and certified. In this paper, we propose the Conversational Logic Injected Neural Network (CLINN). This novel network merges dialog managers “programmed” using logical rules and a Sequenceto-Sequence Neural Network. We experimented with the Restaurant topic of the MultiWOZ dataset. Results show that injected rules are effective when training data set are scarce as well as when more data are available.
ArXiv, 2019
Intent Detection and Slot Filling are two pillar tasks in Spoken Natural Language Understanding. ... more Intent Detection and Slot Filling are two pillar tasks in Spoken Natural Language Understanding. Common approaches adopt joint Deep Learning architectures in attention-based recurrent frameworks. In this work, we aim at exploiting the success of "recurrence-less" models for these tasks. We introduce Bert-Joint, i.e., a multi-lingual joint text classification and sequence labeling framework. The experimental evaluation over two well-known English benchmarks demonstrates the strong performances that can be obtained with this model, even when few annotated data is available. Moreover, we annotated a new dataset for the Italian language, and we observed similar performances without the need for changing the model.
In this paper, we propose a Transfer Learning technique for Named Entity Recognition that is able... more In this paper, we propose a Transfer Learning technique for Named Entity Recognition that is able to flexibly deal with domain changes. The proposed technique is able to manage both the case when the set of named entities does not change and the case when the set of named entities changes in the target domain. In particular, we focus on the case when the target data contains only the annotation of a target named entity, and the source data is no longer available for the target task. Our solution consists in transferring the parameters from a source model, which are then fine-tuned with the target data. The model architecture is modified when recognizing a new category by adding properly new neurons to the model. Our experiments show that it is possible to effectively transfer learned parameters in both the scenarios, resulting in strong performances over the target categories without degrading the performances on the other named entities.
This paper aims at describing, from an industrial perspective, the experience in delivering conve... more This paper aims at describing, from an industrial perspective, the experience in delivering conversational agents via the development of Iride, a platform able to deploy multi-language task-oriented dialog systems. It has been implemented a set of functionalities that can be aggregated in different ways, in order to build domain independent conversational systems, which are able to satisfy needs of real business cases. Along with algorithms and techniques for end to end Dialog management, such as Natural Language Understanding (NLU), Question Answering (QA) and Dialog State tracking and policy management, the technical insights leveraged into the platform are described by outlining the requirements and constraints emerging from these on the field experiences.1
ArXiv, 2019
The widespread use of conversational and question answering systems made it necessary to improve ... more The widespread use of conversational and question answering systems made it necessary to improve the performances of speaker intent detection and understanding of related semantic slots, i.e., Spoken Language Understanding (SLU). Often, these tasks are approached with supervised learning methods, which needs considerable labeled datasets. This paper presents the first Italian dataset for SLU. It is derived through a semi-automatic procedure and is used as a benchmark of various open source and commercial systems.
Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020, 2020
Transfer learning has been proven to be effective, especially when data for the target domain/tas... more Transfer learning has been proven to be effective, especially when data for the target domain/task is scarce. Sometimes data for a similar task is only available in another language because it may be very specific. In this paper, we explore the use of machine-translated data to transfer models on a related domain. Specifically, we transfer models from the question duplication task (QDT) to similar FAQ selection tasks. The source domain is the wellknown English Quora dataset, while the target domain is a collection of small Italian datasets for real case scenarios consisting of FAQ groups retrieved by pivoting on common answers. Our results show great improvements in the zero-shot learning setting and modest improvements using the standard transfer approach for direct in-domain adaptation 1 .
Proceedings. Fourteenth International Conference on Pattern Recognition (Cat. No.98EX170)
An image processing method for features extraction and segmentation from three-dimensional (3D) i... more An image processing method for features extraction and segmentation from three-dimensional (3D) image datasets is presented. Kohonen's self-organizing map (SOM) is used to perform segmentation. Previously, the segmentation method worked on a 2D dataset based on a projection of the three-dimensional dataset (Nguyen et al., 1998). Our 3D approach to segment biological images preserves the 3D object orientations with respect
Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP '98 (Cat. No.98CH36181), 1998
We present a computer vision system based on an integrated neural network architecture. In the lo... more We present a computer vision system based on an integrated neural network architecture. In the low level vision subsystem, a network of networks-a biologically inspired network is used to recursively perform ltering, segmentation and edge detection; in the intermediate level and the high level, hierarchically structured arrays of self-organizing tree maps-extension of the popular self-organizing map are utilized to carry out image feature analysis. The system has been applied to solve a n umber of real world problems. Some interesting and encouraging results will be reported.
Three-Dimensional and Multidimensional Microscopy: Image Acquisition and Processing V, 1998
ABSTRACT Imaging thick specimens in 3D transmission confocal modes presents two key problems. The... more ABSTRACT Imaging thick specimens in 3D transmission confocal modes presents two key problems. The first problem is variable aberrations introduced by changes in refractive index. The second problem is revealed when visualizing acquired data, where thick 3D datasets are difficult to interpret. In this paper we present our emerging solutions to these problems. Aberrations can be classified as simple tip-tilt deflection of the beam, or more complicated higher order aberrations. We discuss our results which demonstrate successful on-the- fly detection and correction for tip-tilt. For detecting higher order aberrations, we have chosen to investigate the wavefront curvature sensing technique. The second problem of rendering thick 3D datasets can be solved by extracting features of interest from the background. Simple intensity thresholding is not sufficient for complex biological specimens. And image processing in only 2D neglects any 3D structure. Use of Kohonen&#39;s self-organizing map neural network in 3D results in clear segmentation of features for sample chromosome specimens.
Uploads
Papers by Raniero Romagnoli