Large-scale auto-regressive language models pretrained on massive text have demonstrated their impressive ability to perform new natural language tasks with only a few text examples, without the need for fine-tuning. Recent studies further show that such a few-shot learning ability can be extended to the text-image setting by training an encoder to encode the images into embeddings functioning like the text embeddings of the language model. Interested in exploring the possibility of transferring the few-shot learning ability to the audio-text setting, we propose a novel speech understanding framework, WavPrompt, where we finetune a wav2vec model to generate a sequence of audio embeddings understood by the language model. We show that WavPrompt is a few-shot learner that can perform speech understanding tasks better than a naive text baseline. We conduct detailed ablation studies on different components and hyperparameters to empirically identify the best model configuration. In addition, we conduct a non-speech understanding experiment to show WavPrompt can extract more information than just the transcriptions. Code is available at https://github.com/Hertin/WavPrompt
The finite invert Beta-Liouville mixture model (IBLMM) has recently gained some attention due to its positive data modeling capability. Under the conventional variational inference (VI) framework, the analytically tractable solution to the optimization of the variational posterior distribution cannot be obtained, since the variational object function involves evaluation of intractable moments. With the recently proposed extended variational inference (EVI) framework, a new function is proposed to replace the original variational object function in order to avoid intractable moment computation, so that the analytically tractable solution of the IBLMM can be derived in an elegant way. The good performance of the proposed approach is demonstrated by experiments with both synthesized data and a real-world application namely text categorization.
The task of learning from only a few examples (called a few-shot setting) is of key importance and relevance to a real-world setting. For question answering (QA), the current state-of-the-art pre-trained models typically need fine-tuning on tens of thousands of examples to obtain good results. Their performance degrades significantly in a few-shot setting (< 100 examples). To address this, we propose a simple fine-tuning framework that leverages pre-trained text-to-text models and is directly aligned with their pre-training framework. Specifically, we construct the input as a concatenation of the question, a mask token representing the answer span and a context. Given this input, the model is fine-tuned using the same objective as that of its pre-training objective. Through experimental studies on various few-shot configurations, we show that this formulation leads to significant gains on multiple QA benchmarks (an absolute gain of 34.2 F1 points on average when there are only 16 training examples). The gains extend further when used with larger models (Eg:- 72.3 F1 on SQuAD using BART-large with only 32 examples) and translate well to a multilingual setting . On the multilingual TydiQA benchmark, our model outperforms the XLM-Roberta-large by an absolute margin of upto 40 F1 points and an average of 33 F1 points in a few-shot setting (<= 64 training examples). We conduct detailed ablation studies to analyze factors contributing to these gains.
In this work we present a systematic empirical study focused on the suitability of the state-of-the-art multilingual encoders for cross-lingual document and sentence retrieval tasks across a number of diverse language pairs. We first treat these models as multilingual text encoders and benchmark their performance in unsupervised ad-hoc sentence- and document-level CLIR. In contrast to supervised language understanding, our results indicate that for unsupervised document-level CLIR -- a setup with no relevance judgments for IR-specific fine-tuning -- pretrained multilingual encoders on average fail to significantly outperform earlier models based on CLWEs. For sentence-level retrieval, we do obtain state-of-the-art performance: the peak scores, however, are met by multilingual encoders that have been further specialized, in a supervised fashion, for sentence understanding tasks, rather than using their vanilla 'off-the-shelf' variants. Following these results, we introduce localized relevance matching for document-level CLIR, where we independently score a query against document sections. In the second part, we evaluate multilingual encoders fine-tuned in a supervised fashion (i.e., we learn to rank) on English relevance data in a series of zero-shot language and domain transfer CLIR experiments. Our results show that supervised re-ranking rarely improves the performance of multilingual transformers as unsupervised base rankers. Finally, only with in-domain contrastive fine-tuning (i.e., same domain, only language transfer), we manage to improve the ranking quality. We uncover substantial empirical differences between cross-lingual retrieval results and results of (zero-shot) cross-lingual transfer for monolingual retrieval in target languages, which point to "monolingual overfitting" of retrieval models trained on monolingual data.
In human speech, the attitude of a speaker cannot be fully expressed only by the textual content. It has to come along with the intonation. Declarative questions are commonly used in daily Cantonese conversations, and they are usually uttered with rising intonation. Vanilla neural text-to-speech (TTS) systems are not capable of synthesizing rising intonation for these sentences due to the loss of semantic information. Though it has become more common to complement the systems with extra language models, their performance in modeling rising intonation is not well studied. In this paper, we propose to complement the Cantonese TTS model with a BERT-based statement/question classifier. We design different training strategies and compare their performance. We conduct our experiments on a Cantonese corpus named CanTTS. Empirical results show that the separate training approach obtains the best generalization performance and feasibility.
Speech transcription, emotion recognition, and language identification are usually considered to be three different tasks. Each one requires a different model with a different architecture and training process. We propose using a recurrent neural network transducer (RNN-T)-based speech-to-text (STT) system as a common component that can be used for emotion recognition and language identification as well as for speech recognition. Our work extends the STT system for emotion classification through minimal changes, and shows successful results on the IEMOCAP and MELD datasets. In addition, we demonstrate that by adding a lightweight component to the RNN-T module, it can also be used for language identification. In our evaluations, this new classifier demonstrates state-of-the-art accuracy for the NIST-LRE-07 dataset.
The remarkable performance gains realized by large pretrained models, e.g., GPT-3, hinge on the massive amounts of data they are exposed to during training. Analogously, distilling such large models to compact models for efficient deployment also necessitates a large amount of (labeled or unlabeled) training data. In this paper, we propose the teacher-guided training (TGT) framework for training a high-quality compact model that leverages the knowledge acquired by pretrained generative models, while obviating the need to go through a large volume of data. TGT exploits the fact that the teacher has acquired a good representation of the underlying data domain, which typically corresponds to a much lower dimensional manifold than the input space. Furthermore, we can use the teacher to explore input space more efficiently through sampling or gradient-based methods; thus, making TGT especially attractive for limited data or long-tail settings. We formally capture this benefit of proposed data-domain exploration in our generalization bounds. We find that TGT can improve accuracy on several image classification benchmarks as well as a range of text classification and retrieval tasks.
Recent advances in multimodal training use textual descriptions to significantly enhance machine understanding of images and videos. Yet, it remains unclear to what extent language can fully capture sensory experiences across different modalities. A well-established approach for characterizing sensory experiences relies on similarity judgments, namely, the degree to which people perceive two distinct stimuli as similar. We explore the relation between human similarity judgments and language in a series of large-scale behavioral studies ($N=1,823$ participants) across three modalities (images, audio, and video) and two types of text descriptors: simple word tags and free-text captions. In doing so, we introduce a novel adaptive pipeline for tag mining that is both efficient and domain-general. We show that our prediction pipeline based on text descriptors exhibits excellent performance, and we compare it against a comprehensive array of 611 baseline models based on vision-, audio-, and video-processing architectures. We further show that the degree to which textual descriptors and models predict human similarity varies across and within modalities. Taken together, these studies illustrate the value of integrating machine learning and cognitive science approaches to better understand the similarities and differences between human and machine representations. We present an interactive visualization at https://words-are-all-you-need.s3.amazonaws.com/index.html for exploring the similarity between stimuli as experienced by humans and different methods reported in the paper.
Aspect-based Sentiment Analysis (ABSA) helps to explain customers' opinions towards products and services. In the past, ABSA models were discriminative, but more recently generative models have been used to generate aspects and polarities directly from text. In contrast, discriminative models commonly first select aspects from the text, and then classify the aspect's polarity. Previous results showed that generative models outperform discriminative models on several English ABSA datasets. Here, we evaluate and contrast two state-of-the-art discriminative and generative models in several settings: cross-lingual, cross-domain, and cross-lingual and domain, to understand generalizability in settings other than English mono-lingual in-domain. Our more thorough evaluation shows that, contrary to previous studies, discriminative models can still outperform generative models in almost all settings.
The vast majority of existing methods and systems for causal inference assume that all variables under consideration are categorical or numerical (e.g., gender, price, blood pressure, enrollment). In this paper, we present CausalNLP, a toolkit for inferring causality from observational data that includes text in addition to traditional numerical and categorical variables. CausalNLP employs the use of meta-learners for treatment effect estimation and supports using raw text and its linguistic properties as both a treatment and a "controlled-for" variable (e.g., confounder). The library is open-source and available at: https://github.com/amaiya/causalnlp.