Abstract Despite the progress made in recent years in addressing natural language understanding (NLU) challenges, the majority of this progress remains to be concentrated on resource-rich languages like English. This work focuses on Persian language, one of the widely spoken languages in the world, and yet there are few NLU datasets available for this language. The availability of high-quality evaluation datasets is a necessity for reliable assessment of the progress on different NLU tasks and domains. We introduce ParsiNLU, the first benchmark in Persian language that includes a range of language understanding tasks—reading comprehension, textual entailment, and so on. These datasets are collected in a multitude of ways, often involving manual annotations by native speakers. This results in over 14.5k new instances across 6 distinct NLU tasks. Additionally, we present the first results on state-of-the-art monolingual and multilingual pre-trained language models on this benchmark and compare them with human performance, which provides valuable insights into our ability to tackle natural language understanding challenges in Persian. We hope ParsiNLU fosters further research and advances in Persian language understanding.1

Past work on acquisition of word-object associations in robots has focused on either fast instruction-based methods which accept highly constrained input or gradual cross-situational learning methods, but not a mixture of both. In this paper, we present an integrated robotic system which allows for a combination of these methods to contribute to the task of learning the labels of objects in AI agents. We demonstrate the expanded word learning capabilities in the outcome system and how learning from both human-human and human-robot dialogues can be achieved in one integrated system.

It has been suggested that early human word learning occurs across learning situations and is bootstrapped by syntactic regularities such as word order. Simulation results from ideal learners and models assuming prior access to structured syn-tactic and semantic representations suggest that it is possible to jointly acquire word order and meanings and that learning is improved as each language capability bootstraps the other.We first present a probabilistic framework for early syntactic bootstrapping in the absence of advanced structured representations, then we use our framework to study the utility of joint acquisition of word order and word referent and its onset, in a memory-limited incremental model. Comparing learning results in the presence and absence of joint acquisition of word order in different ambiguous contexts, improvement in word order results showed an immediate onset, starting in early trials while being affected by context ambiguity. Improvement in word learning results on the other hand, was hindered in early trials where the acquired word order was imperfect,while being facilitated by word order learning in future trials as the acquired word order improved. Furthermore, our results showed that joint acquisition of word order and word referent facilitates one-shot learning of new words as well as inferring intentions of the speaker in ambiguous contexts.

We present a variation of the incremental and memory-limited algorithm in (Sadeghi et al., 2017) for Bayesian cross-situational word learning and evaluate the model in terms of its functional performance and its sensitivity to input order. We show that the functional performance of our sub-optimal model on corpus data is close to that of its optimal counterpart (Frank et al., 2009), while only the sub-optimal model is capable of predicting the input order effects reported in experimental studies.

Hubel Wiesel models, successful in visual processing algorithms, have only recently been used in conceptual representation. Despite the biological plausibility of a Hubel-Wiesel like architecture for conceptual memory and encouraging preliminary results, there is no implementation of how inputs at each layer of the hierarchy should be integrated for processing by a given module, based on the correlation of the features. In our paper, we propose the input integration framework - a set of operations performed on the inputs to the learning modules of the Hubel Wiesel model of conceptual memory. These operations weight the modules as being general or specific and therefore determine how modules can be correlated when fed to parents in the higher layers of the hierarchy. Parallels from Psychology are drawn to support our proposed framework. Simulation results on benchmark data show that implementing local correlation corresponds to the process of early concept generalization to reveal the broadest coherent distinctions of conceptual patterns. Finally, we applied the improved model iteratively over two sets of data, which resulted in the generation of finer grained categorizations, similar to progressive differentiation. Based on our results, we conclude that the model can be used to explain how humans intuitively fit a hierarchical representation for any kind of data.