Deep generative models are increasingly becoming integral parts of the in silico molecule design pipeline and have dual goals of learning the chemical and structural features that render candidate molecules viable while also being flexible enough to generate novel designs. Specifically, Variational Auto Encoders (VAEs) are generative models in which encoder-decoder network pairs are trained to reconstruct training data distributions in such a way that the latent space of the encoder network is smooth. Therefore, novel candidates can be found by sampling from this latent space. However, the scope of architectures and hyperparameters is vast and choosing the best combination for in silico discovery has important implications for downstream success. Therefore, it is important to develop a principled methodology for distinguishing how well a given generative model is able to learn salient molecular features. In this work, we propose a method for measuring how well the latent space of deep generative models is able to encode structural and chemical features of molecular datasets by correlating latent space metrics with metrics from the field of topological data analysis (TDA). We apply our evaluation methodology to a VAE trained on SMILES strings and show that 3D topology information is consistently encoded throughout the latent space of the model.
Manipulating images of complex scenes to reconstruct, insert and/or remove specific object instances is a challenging task. Complex scenes contain multiple semantics and objects, which are frequently cluttered or ambiguous, thus hampering the performance of inpainting models. Conventional techniques often rely on structural information such as object contours in multi-stage approaches that generate unreliable results and boundaries. In this work, we propose a novel deep learning model to alter a complex urban scene by removing a user-specified portion of the image and coherently inserting a new object (e.g. car or pedestrian) in that scene. Inspired by recent works on image inpainting, our proposed method leverages the semantic segmentation to model the content and structure of the image, and learn the best shape and location of the object to insert. To generate reliable results, we design a new decoder block that combines the semantic segmentation and generation task to guide better the generation of new objects and scenes, which have to be semantically consistent with the image. Our experiments, conducted on two large-scale datasets of urban scenes (Cityscapes and Indian Driving), show that our proposed approach successfully address the problem of semantically-guided inpainting of complex urban scene.
As a concise form of user reviews, tips have unique advantages to explain the search results, assist users' decision making, and further improve user experience in vertical search scenarios. Existing work on tip generation does not take query into consideration, which limits the impact of tips in search scenarios. To address this issue, this paper proposes a query-aware tip generation framework, integrating query information into encoding and subsequent decoding processes. Two specific adaptations of Transformer and Recurrent Neural Network (RNN) are proposed. For Transformer, the query impact is incorporated into the self-attention computation of both the encoder and the decoder. As for RNN, the query-aware encoder adopts a selective network to distill query-relevant information from the review, while the query-aware decoder integrates the query information into the attention computation during decoding. The framework consistently outperforms the competing methods on both public and real-world industrial datasets. Last but not least, online deployment experiments on Dianping demonstrate the advantage of the proposed framework for tip generation as well as its online business values.
Non-autoregressive models generate target words in a parallel way, which achieve a faster decoding speed but at the sacrifice of translation accuracy. To remedy a flawed translation by non-autoregressive models, a promising approach is to train a conditional masked translation model (CMTM), and refine the generated results within several iterations. Unfortunately, such approach hardly considers the \textit{sequential dependency} among target words, which inevitably results in a translation degradation. Hence, instead of solely training a Transformer-based CMTM, we propose a Self-Review Mechanism to infuse sequential information into it. Concretely, we insert a left-to-right mask to the same decoder of CMTM, and then induce it to autoregressively review whether each generated word from CMTM is supposed to be replaced or kept. The experimental results (WMT14 En$\leftrightarrow$De and WMT16 En$\leftrightarrow$Ro) demonstrate that our model uses dramatically less training computations than the typical CMTM, as well as outperforms several state-of-the-art non-autoregressive models by over 1 BLEU. Through knowledge distillation, our model even surpasses a typical left-to-right Transformer model, while significantly speeding up decoding.
Information visualizations such as bar charts and line charts are very popular for exploring data and communicating insights. Interpreting and making sense of such visualizations can be challenging for some people, such as those who are visually impaired or have low visualization literacy. In this work, we introduce a new dataset and present a neural model for automatically generating natural language summaries for charts. The generated summaries provide an interpretation of the chart and convey the key insights found within that chart. Our neural model is developed by extending the state-of-the-art model for the data-to-text generation task, which utilizes a transformer-based encoder-decoder architecture. We found that our approach outperforms the base model on a content selection metric by a wide margin (55.42% vs. 8.49%) and generates more informative, concise, and coherent summaries.
We propose a method for fusing stereo disparity estimation with movement-induced prior information. Instead of independent inference frame-by-frame, we formulate the problem as a non-parametric learning task in terms of a temporal Gaussian process prior with a movement-driven kernel for inter-frame reasoning. We present a hierarchy of three Gaussian process kernels depending on the availability of motion information, where our main focus is on a new gyroscope-driven kernel for handheld devices with low-quality MEMS sensors, thus also relaxing the requirement of having full 6D camera poses available. We show how our method can be combined with two state-of-the-art deep stereo methods. The method either work in a plug-and-play fashion with pre-trained deep stereo networks, or further improved by jointly training the kernels together with encoder-decoder architectures, leading to consistent improvement.
This work presents our ongoing research of unsupervised pretraining in neural machine translation (NMT). In our method, we initialize the weights of the encoder and decoder with two language models that are trained with monolingual data and then fine-tune the model on parallel data using Elastic Weight Consolidation (EWC) to avoid forgetting of the original language modeling tasks. We compare the regularization by EWC with the previous work that focuses on regularization by language modeling objectives. The positive result is that using EWC with the decoder achieves BLEU scores similar to the previous work. However, the model converges 2-3 times faster and does not require the original unlabeled training data during the fine-tuning stage. In contrast, the regularization using EWC is less effective if the original and new tasks are not closely related. We show that initializing the bidirectional NMT encoder with a left-to-right language model and forcing the model to remember the original left-to-right language modeling task limits the learning capacity of the encoder for the whole bidirectional context.
In this work, we use a span-based approach for Vietnamese constituency parsing. Our method follows the self-attention encoder architecture and a chart decoder using a CKY-style inference algorithm. We present analyses of the experiment results of the comparison of our empirical method using pre-training models XLM-Roberta and PhoBERT on both Vietnamese datasets VietTreebank and NIIVTB1. The results show that our model with XLM-Roberta archived the significantly F1-score better than other pre-training models, VietTreebank at 81.19% and NIIVTB1 at 85.70%.
The general aim of multi-focus image fusion is to gather focused regions of different images to generate a unique all-in-focus fused image. Deep learning based methods become the mainstream of image fusion by virtue of its powerful feature representation ability. However, most of the existing deep learning structures failed to balance fusion quality and end-to-end implementation convenience. End-to-end decoder design often leads to unrealistic result because of its non-linear mapping mechanism. On the other hand, generating an intermediate decision map achieves better quality for the fused image, but relies on the rectification with empirical post-processing parameter choices. In this work, to handle the requirements of both output image quality and comprehensive simplicity of structure implementation, we propose a cascade network to simultaneously generate decision map and fused result with an end-to-end training procedure. It avoids the dependence on empirical post-processing methods in the inference stage. To improve the fusion quality, we introduce a gradient aware loss function to preserve gradient information in output fused image. In addition, we design a decision calibration strategy to decrease the time consumption in the application of multiple images fusion. Extensive experiments are conducted to compare with 19 different state-of-the-art multi-focus image fusion structures with 6 assessment metrics. The results prove that our designed structure can generally ameliorate the output fused image quality, while implementation efficiency increases over 30\% for multiple images fusion.
Information in speech signals is not evenly distributed, making it an additional challenge for end-to-end (E2E) speech translation (ST) to learn to focus on informative features. In this paper, we propose adaptive feature selection (AFS) for encoder-decoder based E2E ST. We first pre-train an ASR encoder and apply AFS to dynamically estimate the importance of each encoded speech feature to SR. A ST encoder, stacked on top of the ASR encoder, then receives the filtered features from the (frozen) ASR encoder. We take L0DROP (Zhang et al., 2020) as the backbone for AFS, and adapt it to sparsify speech features with respect to both temporal and feature dimensions. Results on LibriSpeech En-Fr and MuST-C benchmarks show that AFS facilitates learning of ST by pruning out ~84% temporal features, yielding an average translation gain of ~1.3-1.6 BLEU and a decoding speedup of ~1.4x. In particular, AFS reduces the performance gap compared to the cascade baseline, and outperforms it on LibriSpeech En-Fr with a BLEU score of 18.56 (without data augmentation)
Copy mechanism allows sequence-to-sequence models to choose words from the input and put them directly into the output, which is finding increasing use in abstractive summarization. However, since there is no explicit delimiter in Chinese sentences, most existing models for Chinese abstractive summarization can only perform character copy, resulting in inefficient. To solve this problem, we propose a lexicon-constrained copying network that models multi-granularity in both encoder and decoder. On the source side, words and characters are aggregated into the same input memory using a Transformerbased encoder. On the target side, the decoder can copy either a character or a multi-character word at each time step, and the decoding process is guided by a word-enhanced search algorithm that facilitates the parallel computation and encourages the model to copy more words. Moreover, we adopt a word selector to integrate keyword information. Experiments results on a Chinese social media dataset show that our model can work standalone or with the word selector. Both forms can outperform previous character-based models and achieve competitive performances.
Video generation models often operate under the assumption of fixed frame rates, which leads to suboptimal performance when it comes to handling flexible frame rates (e.g., increasing the frame rate of the more dynamic portion of the video as well as handling missing video frames). To resolve the restricted nature of existing video generation models' ability to handle arbitrary timesteps, we propose continuous-time video generation by combining neural ODE (Vid-ODE) with pixel-level video processing techniques. Using ODE-ConvGRU as an encoder, a convolutional version of the recently proposed neural ODE, which enables us to learn continuous-time dynamics, Vid-ODE can learn the spatio-temporal dynamics of input videos of flexible frame rates. The decoder integrates the learned dynamics function to synthesize video frames at any given timesteps, where the pixel-level composition technique is used to maintain the sharpness of individual frames. With extensive experiments on four real-world video datasets, we verify that the proposed Vid-ODE outperforms state-of-the-art approaches under various video generation settings, both within the trained time range (interpolation) and beyond the range (extrapolation). To the best of our knowledge, Vid-ODE is the first work successfully performing continuous-time video generation using real-world videos.
A practical automatic textual math word problems (MWPs) solver should be able to solve various textual MWPs while most existing works only focused on one-unknown linear MWPs. Herein, we propose a simple but efficient method called Universal Expression Tree (UET) to make the first attempt to represent the equations of various MWPs uniformly. Then a semantically-aligned universal tree-structured solver (SAU-Solver) based on an encoder-decoder framework is proposed to resolve multiple types of MWPs in a unified model, benefiting from our UET representation. Our SAU-Solver generates a universal expression tree explicitly by deciding which symbol to generate according to the generated symbols' semantic meanings like human solving MWPs. Besides, our SAU-Solver also includes a novel subtree-level semanticallyaligned regularization to further enforce the semantic constraints and rationality of the generated expression tree by aligning with the contextual information. Finally, to validate the universality of our solver and extend the research boundary of MWPs, we introduce a new challenging Hybrid Math Word Problems dataset (HMWP), consisting of three types of MWPs. Experimental results on several MWPs datasets show that our model can solve universal types of MWPs and outperforms several state-of-the-art models.
In this work, we develop a space-time block code for noncoherent communication using techniques from the field of quantum error correction. We decompose the multiple-input multiple-output (MIMO) channel into operators from quantum mechanics, and design a non-coherent space time code using the quantum stabilizer formalism. We derive an optimal decoder, and analyze the former through a quantum mechanical lens. We compare our approach to a comparable coherent approach and a noncoherent differential approach, achieving comparable or better performance.
Deep networks for stereo matching typically leverage 2D or 3D convolutional encoder-decoder architectures to aggregate cost and regularize the cost volume for accurate disparity estimation. Due to content-insensitive convolutions and down-sampling and up-sampling operations, these cost aggregation mechanisms do not take full advantage of the information available in the images. Disparity maps suffer from over-smoothing near occlusion boundaries, and erroneous predictions in thin structures. In this paper, we show how deep adaptive filtering and differentiable semi-global aggregation can be integrated in existing 2D and 3D convolutional networks for end-to-end stereo matching, leading to improved accuracy. The improvements are due to utilizing RGB information from the images as a signal to dynamically guide the matching process, in addition to being the signal we attempt to match across the images. We show extensive experimental results on the KITTI 2015 and Virtual KITTI 2 datasets comparing four stereo networks (DispNetC, GCNet, PSMNet and GANet) after integrating four adaptive filters (segmentation-aware bilateral filtering, dynamic filtering networks, pixel adaptive convolution and semi-global aggregation) into their architectures. Our code is available at https://github.com/ccj5351/DAFStereoNets.
Meta-learning algorithms adapt quickly to new tasks that are drawn from the same task distribution as the training tasks. The mechanism leading to fast adaptation is the conditioning of a downstream predictive model on the inferred representation of the task's underlying data generative process, or \emph{function}. This \emph{meta-representation}, which is computed from a few observed examples of the underlying function, is learned jointly with the predictive model. In this work, we study the implications of this joint training on the transferability of the meta-representations. Our goal is to learn meta-representations that are robust to noise in the data and facilitate solving a wide range of downstream tasks that share the same underlying functions. To this end, we propose a decoupled encoder-decoder approach to supervised meta-learning, where the encoder is trained with a contrastive objective to find a good representation of the underlying function. In particular, our training scheme is driven by the self-supervision signal indicating whether two sets of examples stem from the same function. Our experiments on a number of synthetic and real-world datasets show that the representations we obtain outperform strong baselines in terms of downstream performance and noise robustness, even when these baselines are trained in an end-to-end manner.
We consider a series of image segmentation methods based on the deep neural networks in order to perform semantic segmentation of electroluminescence (EL) images of thin-film modules. We utilize the encoder-decoder deep neural network architecture. The framework is general such that it can easily be extended to other types of images (e.g. thermography) or solar cell technologies (e.g. crystalline silicon modules). The networks are trained and tested on a sample of images from a database with 6000 EL images of Copper Indium Gallium Diselenide (CIGS) thin film modules. We selected two types of features to extract, shunts and so called "droplets". The latter feature is often observed in the set of images. Several models are tested using various combinations of encoder-decoder layers, and a procedure is proposed to select the best model. We show exemplary results with the best selected model. Furthermore, we applied the best model to the full set of 6000 images and demonstrate that the automated segmentation of EL images can reveal many subtle features which cannot be inferred from studying a small sample of images. We believe these features can contribute to process optimization and quality control.
In contemporary society, surveillance anomaly detection, i.e., spotting anomalous events such as crimes or accidents in surveillance videos, is a critical task. As anomalies occur rarely, most training data consists of unlabeled videos without anomalous events, which makes the task challenging. Most existing methods use an autoencoder (AE) to learn to reconstruct normal videos; they then detect anomalies based on their failure to reconstruct the appearance of abnormal scenes. However, because anomalies are distinguished by appearance as well as motion, many previous approaches have explicitly separated appearance and motion information-for example, using a pre-trained optical flow model. This explicit separation restricts reciprocal representation capabilities between two types of information. In contrast, we propose an implicit two-path AE (ITAE), a structure in which two encoders implicitly model appearance and motion features, along with a single decoder that combines them to learn normal video patterns. For the complex distribution of normal scenes, we suggest normal density estimation of ITAE features through normalizing flow (NF)-based generative models to learn the tractable likelihoods and identify anomalies using out of distribution detection. NF models intensify ITAE performance by learning normality through implicitly learned features. Finally, we demonstrate the effectiveness of ITAE and its feature distribution modeling on six benchmarks, including databases that contain various anomalies in real-world scenarios.
Automated detection of curvilinear structures, e.g., blood vessels or nerve fibres, from medical and biomedical images is a crucial early step in automatic image interpretation associated to the management of many diseases. Precise measurement of the morphological changes of these curvilinear organ structures informs clinicians for understanding the mechanism, diagnosis, and treatment of e.g. cardiovascular, kidney, eye, lung, and neurological conditions. In this work, we propose a generic and unified convolution neural network for the segmentation of curvilinear structures and illustrate in several 2D/3D medical imaging modalities. We introduce a new curvilinear structure segmentation network (CS2-Net), which includes a self-attention mechanism in the encoder and decoder to learn rich hierarchical representations of curvilinear structures. Two types of attention modules - spatial attention and channel attention - are utilized to enhance the inter-class discrimination and intra-class responsiveness, to further integrate local features with their global dependencies and normalization, adaptively. Furthermore, to facilitate the segmentation of curvilinear structures in medical images, we employ a 1x3 and a 3x1 convolutional kernel to capture boundary features. ...
An advantage of seq2seq abstractive summarization models is that they generate text in a free-form manner, but this flexibility makes it difficult to interpret model behavior. In this work, we analyze summarization decoders in both blackbox and whitebox ways by studying on the entropy, or uncertainty, of the model's token-level predictions. For two strong pre-trained models, PEGASUS and BART on two summarization datasets, we find a strong correlation between low prediction entropy and where the model copies tokens rather than generating novel text. The decoder's uncertainty also connects to factors like sentence position and syntactic distance between adjacent pairs of tokens, giving a sense of what factors make a context particularly selective for the model's next output token. Finally, we study the relationship of decoder uncertainty and attention behavior to understand how attention gives rise to these observed effects in the model. We show that uncertainty is a useful perspective for analyzing summarization and text generation models more broadly.