The paper titled “GAN-BERT: Generative Adversarial Learning for Robust Text Classification with a Bunch of Labeled Examples” has been accepted at ACL2020! It is the result of a recent collaboration between our group and Amazon Seattle.
Here the paper abstract:
Recently, Transformer-based architectures, e.g., BERT, provide impressive results in many Natural Language Processing tasks. However, most of the adopted benchmarks are made of (sometimes hundreds of) thousands examples. In many real scenarios, obtaining high-quality annotated data is expensive and time-consuming; in contrast, unlabeled examples characterizing the target task can be in general easily collected.
One promising method to enable semi-supervised learning has been proposed in image processing, based on Semi-Supervised Generative Adversarial Networks.
In this paper, we propose GAN-BERT that extends the fine-tuning of BERT-like architectures with unlabeled data in a generative adversarial setting. Experimental results show that the requirement for annotated examples can be drastically reduced (up to only 50-100 annotated examples), still obtaining good performances in several sentence classification tasks.
HuRIC (Human Robot Interaction Corpus) is a resource that has been gathered as a collaboration between the Semantic Analytics Group (SAG) from the University of Roma, Tor Vergata, and the Laboratory of Cognitive Cooperating Robots (Lab.Ro.Co.Co.) at Sapienza, University of Rome. The basic idea of this project is to build a corpus for Human Robot Interaction in Natural Language containing information that are yet oriented to a specific application domain, e.g. the house service robotics, but at the same time inspired by sound linguistic theories, that are by definition decoupled from such a domain.
HuRIC 2.0 is available at https://github.com/crux82/huric
Call for Participation SemEval 2020 Task 12
OffensEval – Multilingual Offensive Language Identification in Social Media
“Actionable ethics through Multitask Neural Learning”
authored by D. Rossini, D. Croce, S. Mancini, M. Pellegrino e R. Basili has been accepted for publication at AAAI 2020, and will be presented in New York, next February!
SQuAD-it – A large scale dataset for Question Answering in Italian is now available.
SQuAD-it is derived from the SQuAD dataset and it is obtained through the semi-automatic translation of the SQuAD dataset in Italian. It represents a large-scale dataset for open question answering processes on factoid questions in Italian. This dataset contains more than 60,000 question/answer pairs derived from the original English dataset.
You can download it at: https://github.com/crux82/squad-it
The paper “Deep Learning in Semantic Kernel Spaces” authored by D. Croce, S. Filice and R. Basili has been accepted for publication at ACL 2017!
“A Discriminative Approach to Grounded Natural Language Learning in Interactive Robotics”
authored by Emanuele Bastianelli, Danilo Croce, Andrea Vanzo, Roberto Basili and Daniele Nardi,
has been accepted at IJCAI 2016 as a Full Paper. Given the 2,294 submissions, acceptance rate at IJCAI 2016 has been 25%.
The SAG group has contributed to Maker Faire 2015, in Rome, by supporting the Reveal team in the release of Giulia, a talking avatar. At stand J8 Giulia can talk with users about “L’eleganza del Cibo“, an exhibition, supported by the Gattinoni firm, about the Italian way to fashion and food. The exhibition is held at the Marcati di Traiano, in Rome in May-October 2015.
More details at: UniIndustria: Softlab & Reveal @ Maker Faire 2015
Pisa, 10-12 December 2014
The thirteenth Symposium of the Italian Association for Artificial Intelligence (AI*IA) will take place in Pisa, Italy, from 10 to 12 December 2014. It will be hosted by the Department of Computer Science at the University of Pisa.
CLiC-it, is the 1st Italian Conference on Computational Linguistics aiming to establish a reference forum for research on Computational Linguistics of the Italian community. CLiC-it 2014 will take place in Pisa from 9-10 December 2014. The Italian Computational Linguistics Conference, will be co-located with the XIII Symposium of the Italian Association for Artificial Intelligence (AI*IA 2014), including the worskshop EVALITA 2014.