Align after Pre-train: Improving Multilingual Generative Models with Cross-lingual Alignment

Multilingual generative models obtain remarkable cross-lingual capabilities through pre-training on large-scale corpora. However, they still exhibit a performance bias toward high-resource languages, and learn isolated distributions of sentence representations across languages. To bridge this gap, we propose a simple yet effective alignment framework exploiting pairs of translation sentences. It aligns the internal sentence representations across different languages via multilingual contrastive learning and aligns model outputs by answering prompts in different languages. Experimental results demonstrate that even with less than 0.1 {\textperthousand} of pre-training tokens, our alignment framework significantly boosts the cross-lingual abilities of generative models and mitigates the performance gap. Further analysis reveals that it results in a better internal multilingual representation distribution of multilingual models.

Interpreting and Exploiting Functional Specialization in Multi-Head Attention under Multi-task Learning

Transformer-based models, even though achieving super-human performance on several downstream tasks, are often regarded as a black box and used as a whole. It is still unclear what mechanisms they have learned, especially their core module: multi-head attention. Inspired by functional specialization in the human brain, which helps to efficiently handle multiple tasks, this work attempts to figure out whether the multi-head attention module will evolve similar function separation under multi-tasking training. If it is, can this mechanism further improve the model performance? To investigate these questions, we introduce an interpreting method to quantify the degree of functional specialization in multi-head attention. We further propose a simple multi-task training method to increase functional specialization and mitigate negative information transfer in multi-task learning. Experimental results on seven pre-trained transformer models have demonstrated that multi-head attention does evolve functional specialization phenomenon after multi-task training which is affected by the similarity of tasks. Moreover, the multi-task training strategy based on functional specialization boosts performance in both multi-task learning and transfer learning without adding any parameters.

Contrast, Attend and Diffuse to Decode High-Resolution Images from Brain Activities

Decoding visual stimuli from neural responses recorded by functional Magnetic Resonance Imaging (fMRI) presents an intriguing intersection between cognitive neuroscience and machine learning, promising advancements in understanding human visual perception and building non-invasive brain-machine interfaces. However, the task is challenging due to the noisy nature of fMRI signals and the intricate pattern of brain visual representations. To mitigate these challenges, we introduce a two-phase fMRI representation learning framework. The first phase pre-trains an fMRI feature learner with a proposed Double-contrastive Mask Auto-encoder to learn denoised representations. The second phase tunes the feature learner to attend to neural activation patterns most informative for visual reconstruction with guidance from an image auto-encoder. The optimized fMRI feature learner then conditions a latent diffusion model to reconstruct image stimuli from brain activities. Experimental results demonstrate our model's superiority in generating high-resolution and semantically accurate images, substantially exceeding previous state-of-the-art methods by 39.34% in the 50-way-top-1 semantic classification accuracy. Our research invites further exploration of the decoding task's potential and contributes to the development of non-invasive brain-machine interfaces.

Language Cognition and Language Computation -- Human and Machine Language Understanding

Language understanding is a key scientific issue in the fields of cognitive and computer science. However, the two disciplines differ substantially in the specific research questions. Cognitive science focuses on analyzing the specific mechanism of the brain and investigating the brain's response to language; few studies have examined the brain's language system as a whole. By contrast, computer scientists focus on the efficiency of practical applications when choosing research questions but may ignore the most essential laws of language. Given these differences, can a combination of the disciplines offer new insights for building intelligent language models and studying language cognitive mechanisms? In the following text, we first review the research questions, history, and methods of language understanding in cognitive and computer science, focusing on the current progress and challenges. We then compare and contrast the research of language understanding in cognitive and computer sciences. Finally, we review existing work that combines insights from language cognition and language computation and offer prospects for future development trends.

Improved Target-specific Stance Detection on Social Media Platforms by Delving into Conversation Threads

Target-specific stance detection on social media, which aims at classifying a textual data instance such as a post or a comment into a stance class of a target issue, has become an emerging opinion mining paradigm of importance. An example application would be to overcome vaccine hesitancy in combating the coronavirus pandemic. However, existing stance detection strategies rely merely on the individual instances which cannot always capture the expressed stance of a given target. In response, we address a new task called conversational stance detection which is to infer the stance towards a given target (e.g., COVID-19 vaccination) when given a data instance and its corresponding conversation thread. To tackle the task, we first propose a benchmarking conversational stance detection (CSD) dataset with annotations of stances and the structures of conversation threads among the instances based on six major social media platforms in Hong Kong. To infer the desired stances from both data instances and conversation threads, we propose a model called Branch-BERT that incorporates contextual information in conversation threads. Extensive experiments on our CSD dataset show that our proposed model outperforms all the baseline models that do not make use of contextual information. Specifically, it improves the F1 score by 10.3% compared with the state-of-the-art method in the SemEval-2016 Task 6 competition. This shows the potential of incorporating rich contextual information on detecting target-specific stances on social media platforms and implies a more practical way to construct future stance detection tasks.

NCLS: Neural Cross-Lingual Summarization

Cross-lingual summarization (CLS) is the task to produce a summary in one particular language for a source document in a different language. Existing methods simply divide this task into two steps: summarization and translation, leading to the problem of error propagation. To handle that, we present an end-to-end CLS framework, which we refer to as Neural Cross-Lingual Summarization (NCLS), for the first time. Moreover, we propose to further improve NCLS by incorporating two related tasks, monolingual summarization and machine translation, into the training process of CLS under multi-task learning. Due to the lack of supervised CLS data, we propose a round-trip translation strategy to acquire two high-quality large-scale CLS datasets based on existing monolingual summarization datasets. Experimental results have shown that our NCLS achieves remarkable improvement over traditional pipeline methods on both English-to-Chinese and Chinese-to-English CLS human-corrected test sets. In addition, NCLS with multi-task learning can further significantly improve the quality of generated summaries. We make our dataset and code publicly available here: http://www.nlpr.ia.ac.cn/cip/dataset.htm.

Understanding Memory Modules on Learning Simple Algorithms

Recent work has shown that memory modules are crucial for the generalization ability of neural networks on learning simple algorithms. However, we still have little understanding of the working mechanism of memory modules. To alleviate this problem, we apply a two-step analysis pipeline consisting of first inferring hypothesis about what strategy the model has learned according to visualization and then verify it by a novel proposed qualitative analysis method based on dimension reduction. Using this method, we have analyzed two popular memory-augmented neural networks, neural Turing machine and stack-augmented neural network on two simple algorithm tasks including reversing a random sequence and evaluation of arithmetic expressions. Results have shown that on the former task both models can learn to generalize and on the latter task only the stack-augmented model can do so. We show that different strategies are learned by the models, in which specific categories of input are monitored and different policies are made based on that to change the memory.

Learning Multimodal Word Representation via Dynamic Fusion Methods

Multimodal models have been proven to outperform text-based models on learning semantic word representations. Almost all previous multimodal models typically treat the representations from different modalities equally. However, it is obvious that information from different modalities contributes differently to the meaning of words. This motivates us to build a multimodal model that can dynamically fuse the semantic representations from different modalities according to different types of words. To that end, we propose three novel dynamic fusion methods to assign importance weights to each modality, in which weights are learned under the weak supervision of word association pairs. The extensive experiments have demonstrated that the proposed methods outperform strong unimodal baselines and state-of-the-art multimodal models.

Investigating Inner Properties of Multimodal Representation and Semantic Compositionality with Brain-based Componential Semantics

Multimodal models have been proven to outperform text-based approaches on learning semantic representations. However, it still remains unclear what properties are encoded in multimodal representations, in what aspects do they outperform the single-modality representations, and what happened in the process of semantic compositionality in different input modalities. Considering that multimodal models are originally motivated by human concept representations, we assume that correlating multimodal representations with brain-based semantics would interpret their inner properties to answer the above questions. To that end, we propose simple interpretation methods based on brain-based componential semantics. First we investigate the inner properties of multimodal representations by correlating them with corresponding brain-based property vectors. Then we map the distributed vector space to the interpretable brain-based componential space to explore the inner properties of semantic compositionality. Ultimately, the present paper sheds light on the fundamental questions of natural language understanding, such as how to represent the meaning of words and how to combine word meanings into larger units.

Learning Sentence Representation with Guidance of Human Attention

Recently, much progress has been made in learning general-purpose sentence representations that can be used across domains. However, most of the existing models typically treat each word in a sentence equally. In contrast, extensive studies have proven that human read sentences efficiently by making a sequence of fixation and saccades. This motivates us to improve sentence representations by assigning different weights to the vectors of the component words, which can be treated as an attention mechanism on single sentences. To that end, we propose two novel attention models, in which the attention weights are derived using significant predictors of human reading time, i.e., Surprisal, POS tags and CCG supertags. The extensive experiments demonstrate that the proposed methods significantly improve upon the state-of-the-art sentence representation models.