Human infants acquire their verbal lexicon from minimal prior knowledge of language based on the statistical properties of phonological distributions and the co-occurrence of other sensory stimuli. In this study, we propose a novel fully unsupervised learning method discovering speech units by utilizing phonological information as a distributional cue and object information as a co-occurrence cue. The proposed method can not only (1) acquire words and phonemes from speech signals using unsupervised learning, but can also (2) utilize object information based on multiple modalities (i.e., vision, tactile, and auditory) simultaneously. The proposed method is based on the Nonparametric Bayesian Double Articulation Analyzer (NPB-DAA) discovering phonemes and words from phonological features, and Multimodal Latent Dirichlet Allocation (MLDA) categorizing multimodal information obtained from objects. In the experiment, the proposed method showed higher word discovery performance than the baseline methods. In particular, words that expressed the characteristics of the object (i.e., words corresponding to nouns and adjectives) were segmented accurately. Furthermore, we examined how learning performance is affected by differences in the importance of linguistic information. When the weight of the word modality was increased, the performance was further improved compared to the fixed condition.
Speech emotion recognition (SER) is a crucial research topic in human-computer interactions. Existing works are mainly based on manually designed models. Despite their great success, these methods heavily rely on historical experience, which are time-consuming but cannot exhaust all possible structures. To address this problem, we propose a neural architecture search (NAS) based framework for SER, called "EmotionNAS". We take spectrogram and wav2vec features as the inputs, followed with NAS to optimize the network structure for these features separately. We further incorporate complementary information in these features through decision-level fusion. Experimental results on IEMOCAP demonstrate that our method succeeds over existing state-of-the-art strategies on SER.
While deep learning surpasses human-level performance in narrow and specific vision tasks, it is fragile and over-confident in classification. For example, minor transformations in perspective, illumination, or object deformation in the image space can result in drastically different labeling, which is especially transparent via adversarial perturbations. On the other hand, human visual perception is orders of magnitude more robust to changes in the input stimulus. But unfortunately, we are far from fully understanding and integrating the underlying mechanisms that result in such robust perception. In this work, we introduce a novel method of incorporating temporal dynamics into static image understanding. We describe a neuro-inspired method that decomposes a single image into a series of coarse-to-fine images that simulates how biological vision integrates information over time. Next, we demonstrate how our novel visual perception framework can utilize this information "over time" using a biologically plausible algorithm with recurrent units, and as a result, significantly improving its accuracy and robustness over standard CNNs. We also compare our proposed approach with state-of-the-art models and explicitly quantify our adversarial robustness properties through multiple ablation studies. Our quantitative and qualitative results convincingly demonstrate exciting and transformative improvements over the standard computer vision and deep learning architectures used today.
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
We introduce Contrastive Intrinsic Control (CIC), an algorithm for unsupervised skill discovery that maximizes the mutual information between skills and state transitions. In contrast to most prior approaches, CIC uses a decomposition of the mutual information that explicitly incentivizes diverse behaviors by maximizing state entropy. We derive a novel lower bound estimate for the mutual information which combines a particle estimator for state entropy to generate diverse behaviors and contrastive learning to distill these behaviors into distinct skills. We evaluate our algorithm on the Unsupervised Reinforcement Learning Benchmark, which consists of a long reward-free pre-training phase followed by a short adaptation phase to downstream tasks with extrinsic rewards. We find that CIC substantially improves over prior unsupervised skill discovery methods and outperforms the next leading overall exploration algorithm in terms of downstream task performance.
Optical flow, which captures motion information across frames, is exploited in recent video inpainting methods through propagating pixels along its trajectories. However, the hand-crafted flow-based processes in these methods are applied separately to form the whole inpainting pipeline. Thus, these methods are less efficient and rely heavily on the intermediate results from earlier stages. In this paper, we propose an End-to-End framework for Flow-Guided Video Inpainting (E$^2$FGVI) through elaborately designed three trainable modules, namely, flow completion, feature propagation, and content hallucination modules. The three modules correspond with the three stages of previous flow-based methods but can be jointly optimized, leading to a more efficient and effective inpainting process. Experimental results demonstrate that the proposed method outperforms state-of-the-art methods both qualitatively and quantitatively and shows promising efficiency. The code is available at https://github.com/MCG-NKU/E2FGVI.
Nowadays the measure between heterogeneous data is still an open problem for cross-modal retrieval. The core of cross-modal retrieval is how to measure the similarity between different types of data. Many approaches have been developed to solve the problem. As one of the mainstream, approaches based on subspace learning pay attention to learning a common subspace where the similarity among multi-modal data can be measured directly. However, many of the existing approaches only focus on learning a latent subspace. They ignore the full use of discriminative information so that the semantically structural information is not well preserved. Therefore satisfactory results can not be achieved as expected. We in this paper propose a discriminative supervised subspace learning for cross-modal retrieval(DS2L), to make full use of discriminative information and better preserve the semantically structural information. Specifically, we first construct a shared semantic graph to preserve the semantic structure within each modality. Subsequently, the Hilbert-Schmidt Independence Criterion(HSIC) is introduced to preserve the consistence between feature-similarity and semantic-similarity of samples. Thirdly, we introduce a similarity preservation term, thus our model can compensate for the shortcomings of insufficient use of discriminative data and better preserve the semantically structural information within each modality. The experimental results obtained on three well-known benchmark datasets demonstrate the effectiveness and competitiveness of the proposed method against the compared classic subspace learning approaches.
Shannon entropy is often a quantity of interest to linguists studying the communicative capacity of human language. However, entropy must typically be estimated from observed data because researchers do not have access to the underlying probability distribution that gives rise to these data. While entropy estimation is a well-studied problem in other fields, there is not yet a comprehensive exploration of the efficacy of entropy estimators for use with linguistic data. In this work, we fill this void, studying the empirical effectiveness of different entropy estimators for linguistic distributions. In a replication of two recent information-theoretic linguistic studies, we find evidence that the reported effect size is over-estimated due to over-reliance on poor entropy estimators. Finally, we end our paper with concrete recommendations for entropy estimation depending on distribution type and data availability.
Correctly quantifying the robustness of machine learning models is a central aspect in judging their suitability for specific tasks, and thus, ultimately, for generating trust in the models. We show that the widely used concept of adversarial robustness and closely related metrics based on counterfactuals are not necessarily valid metrics for determining the robustness of ML models against perturbations that occur "naturally", outside specific adversarial attack scenarios. Additionally, we argue that generic robustness metrics in principle are insufficient for determining real-world-robustness. Instead we propose a flexible approach that models possible perturbations in input data individually for each application. This is then combined with a probabilistic approach that computes the likelihood that a real-world perturbation will change a prediction, thus giving quantitative information of the robustness of the trained machine learning model. The method does not require access to the internals of the classifier and thus in principle works for any black-box model. It is, however, based on Monte-Carlo sampling and thus only suited for input spaces with small dimensions. We illustrate our approach on two dataset, as well as on analytically solvable cases. Finally, we discuss ideas on how real-world robustness could be computed or estimated in high-dimensional input spaces.
Visual question answering is a vision-and-language multimodal task, that aims at predicting answers given samples from the question and image modalities. Most recent methods focus on learning a good joint embedding space of images and questions, either by improving the interaction between these two modalities, or by making it a more discriminant space. However, how informative this joint space is, has not been well explored. In this paper, we propose a novel regularization for VQA models, Constrained Optimization using Barlow's theory (COB), that improves the information content of the joint space by minimizing the redundancy. It reduces the correlation between the learned feature components and thereby disentangles semantic concepts. Our model also aligns the joint space with the answer embedding space, where we consider the answer and image+question as two different `views' of what in essence is the same semantic information. We propose a constrained optimization policy to balance the categorical and redundancy minimization forces. When built on the state-of-the-art GGE model, the resulting model improves VQA accuracy by 1.4% and 4% on the VQA-CP v2 and VQA v2 datasets respectively. The model also exhibits better interpretability.