Most current multi-modal summarization methods follow a cascaded manner, where an off-the-shelf object detector is first used to extract visual features, then these features are fused with language representations to generate the summary with an encoder-decoder model. The cascaded way cannot capture the semantic alignments between images and paragraphs, which are crucial to a precise summary. In this paper, we propose ViL-Sum to jointly model paragraph-level \textbf{Vi}sion-\textbf{L}anguage Semantic Alignment and Multi-Modal \textbf{Sum}marization. The core of ViL-Sum is a joint multi-modal encoder with two well-designed tasks, image reordering and image selection. The joint multi-modal encoder captures the interactions between modalities, where the reordering task guides the model to learn paragraph-level semantic alignment and the selection task guides the model to selected summary-related images in the final summary. Experimental results show that our proposed ViL-Sum significantly outperforms current state-of-the-art methods. In further analysis, we find that two well-designed tasks and joint multi-modal encoder can effectively guide the model to learn reasonable paragraphs-images and summary-images relations.
Recently, deep learning methods have achieved state-of-the-art performance in many medical image segmentation tasks. Many of these are based on convolutional neural networks (CNNs). For such methods, the encoder is the key part for global and local information extraction from input images; the extracted features are then passed to the decoder for predicting the segmentations. In contrast, several recent works show a superior performance with the use of transformers, which can better model long-range spatial dependencies and capture low-level details. However, transformer as sole encoder underperforms for some tasks where it cannot efficiently replace the convolution based encoder. In this paper, we propose a model with double encoders for 3D biomedical image segmentation. Our model is a U-shaped CNN augmented with an independent transformer encoder. We fuse the information from the convolutional encoder and the transformer, and pass it to the decoder to obtain the results. We evaluate our methods on three public datasets from three different challenges: BTCV, MoDA and Decathlon. Compared to the state-of-the-art models with and without transformers on each task, our proposed method obtains higher Dice scores across the board.
In this paper, we focus on improving binary 2D instance segmentation to assist humans in labeling ground truth datasets with polygons. Humans labeler just have to draw boxes around objects, and polygons are generated automatically. To be useful, our system has to run on CPUs in real-time. The most usual approach for binary instance segmentation involves encoder-decoder networks. This report evaluates state-of-the-art encoder-decoder networks and proposes a method for improving instance segmentation quality using these networks. Alongside network architecture improvements, our proposed method relies upon providing extra information to the network input, so-called extreme points, i.e. the outermost points on the object silhouette. The user can label them instead of a bounding box almost as quickly. The bounding box can be deduced from the extreme points as well. This method produces better IoU compared to other state-of-the-art encoder-decoder networks and also runs fast enough when it is deployed on a CPU.
Multi-scale features have been proven highly effective for object detection but often come with huge and even prohibitive extra computation costs, especially for the recent Transformer-based detectors. In this paper, we propose Iterative Multi-scale Feature Aggregation (IMFA) -- a generic paradigm that enables efficient use of multi-scale features in Transformer-based object detectors. The core idea is to exploit sparse multi-scale features from just a few crucial locations, and it is achieved with two novel designs. First, IMFA rearranges the Transformer encoder-decoder pipeline so that the encoded features can be iteratively updated based on the detection predictions. Second, IMFA sparsely samples scale-adaptive features for refined detection from just a few keypoint locations under the guidance of prior detection predictions. As a result, the sampled multi-scale features are sparse yet still highly beneficial for object detection. Extensive experiments show that the proposed IMFA boosts the performance of multiple Transformer-based object detectors significantly yet with only slight computational overhead.
Users' search tasks have become increasingly complicated, requiring multiple queries and interactions with the results. Recent studies have demonstrated that modeling the historical user behaviors in a session can help understand the current search intent. Existing context-aware ranking models primarily encode the current session sequence (from the first behavior to the current query) and compute the ranking score using the high-level representations. However, there is usually some noise in the current session sequence (useless behaviors for inferring the search intent) that may affect the quality of the encoded representations. To help the encoding of the current user behavior sequence, we propose to use a decoder and the information of future sequences and a supplemental query. Specifically, we design three generative tasks that can help the encoder to infer the actual search intent: (1) predicting future queries, (2) predicting future clicked documents, and (3) predicting a supplemental query. We jointly learn the ranking task with these generative tasks using an encoder-decoder structured approach. Extensive experiments on two public search logs demonstrate that our model outperforms all existing baselines, and the designed generative tasks can actually help the ranking task. Besides, additional experiments also show that our approach can be easily applied to various Transformer-based encoder-decoder models and improve their performance.
User simulators (USs) are commonly used to train task-oriented dialogue systems (DSs) via reinforcement learning. The interactions often take place on semantic level for efficiency, but there is still a gap from semantic actions to natural language, which causes a mismatch between training and deployment environment. Incorporating a natural language generation (NLG) module with USs during training can partly deal with this problem. However, since the policy and NLG of USs are optimised separately, these simulated user utterances may not be natural enough in a given context. In this work, we propose a generative transformer-based user simulator (GenTUS). GenTUS consists of an encoder-decoder structure, which means it can optimise both the user policy and natural language generation jointly. GenTUS generates both semantic actions and natural language utterances, preserving interpretability and enhancing language variation. In addition, by representing the inputs and outputs as word sequences and by using a large pre-trained language model we can achieve generalisability in feature representation. We evaluate GenTUS with automatic metrics and human evaluation. Our results show that GenTUS generates more natural language and is able to transfer to an unseen ontology in a zero-shot fashion. In addition, its behaviour can be further shaped with reinforcement learning opening the door to training specialised user simulators.
Recent transformer-based offline video instance segmentation (VIS) approaches achieve encouraging results and significantly outperform online approaches. However, their reliance on the whole video and the immense computational complexity caused by full Spatio-temporal attention limit them in real-life applications such as processing lengthy videos. In this paper, we propose a single-stage transformer-based efficient online VIS framework named InstanceFormer, which is especially suitable for long and challenging videos. We propose three novel components to model short-term and long-term dependency and temporal coherence. First, we propagate the representation, location, and semantic information of prior instances to model short-term changes. Second, we propose a novel memory cross-attention in the decoder, which allows the network to look into earlier instances within a certain temporal window. Finally, we employ a temporal contrastive loss to impose coherence in the representation of an instance across all frames. Memory attention and temporal coherence are particularly beneficial to long-range dependency modeling, including challenging scenarios like occlusion. The proposed InstanceFormer outperforms previous online benchmark methods by a large margin across multiple datasets. Most importantly, InstanceFormer surpasses offline approaches for challenging and long datasets such as YouTube-VIS-2021 and OVIS. Code is available at https://github.com/rajatkoner08/InstanceFormer.
For humans, understanding the relationships between objects using visual signals is intuitive. For artificial intelligence, however, this task remains challenging. Researchers have made significant progress studying semantic relationship detection, such as human-object interaction detection and visual relationship detection. We take the study of visual relationships a step further from semantic to geometric. In specific, we predict relative occlusion and relative distance relationships. However, detecting these relationships from a single image is challenging. Enforcing focused attention to task-specific regions plays a critical role in successfully detecting these relationships. In this work, (1) we propose a novel three-decoder architecture as the infrastructure for focused attention; 2) we use the generalized intersection box prediction task to effectively guide our model to focus on occlusion-specific regions; 3) our model achieves a new state-of-the-art performance on distance-aware relationship detection. Specifically, our model increases the distance F1-score from 33.8% to 38.6% and boosts the occlusion F1-score from 34.4% to 41.2%. Our code is publicly available.
Existing top-performance 3D object detectors typically rely on the multi-modal fusion strategy. This design is however fundamentally restricted due to overlooking the modality-specific useful information and finally hampering the model performance. To address this limitation, in this work we introduce a novel modality interaction strategy where individual per-modality representations are learned and maintained throughout for enabling their unique characteristics to be exploited during object detection. To realize this proposed strategy, we design a DeepInteraction architecture characterized by a multi-modal representational interaction encoder and a multi-modal predictive interaction decoder. Experiments on the large-scale nuScenes dataset show that our proposed method surpasses all prior arts often by a large margin. Crucially, our method is ranked at the first position at the highly competitive nuScenes object detection leaderboard.
Accurate facial landmarks are essential prerequisites for many tasks related to human faces. In this paper, an accurate facial landmark detector is proposed based on cascaded transformers. We formulate facial landmark detection as a coordinate regression task such that the model can be trained end-to-end. With self-attention in transformers, our model can inherently exploit the structured relationships between landmarks, which would benefit landmark detection under challenging conditions such as large pose and occlusion. During cascaded refinement, our model is able to extract the most relevant image features around the target landmark for coordinate prediction, based on deformable attention mechanism, thus bringing more accurate alignment. In addition, we propose a novel decoder that refines image features and landmark positions simultaneously. With few parameter increasing, the detection performance improves further. Our model achieves new state-of-the-art performance on several standard facial landmark detection benchmarks, and shows good generalization ability in cross-dataset evaluation.
In this paper, we investigate the problem of string-based molecular generation via variational autoencoders (VAEs) that have served a popular generative approach for various tasks in artificial intelligence. We propose a simple, yet effective idea to improve the performance of VAE for the task. Our main idea is to maintain multiple decoders while sharing a single encoder, i.e., it is a type of ensemble techniques. Here, we first found that training each decoder independently may not be effective as the bias of the ensemble decoder increases severely under its auto-regressive inference. To maintain both small bias and variance of the ensemble model, our proposed technique is two-fold: (a) a different latent variable is sampled for each decoder (from estimated mean and variance offered by the shared encoder) to encourage diverse characteristics of decoders and (b) a collaborative loss is used during training to control the aggregated quality of decoders using different latent variables. In our experiments, the proposed VAE model particularly performs well for generating a sample from out-of-domain distribution.
Dense retrieval (DR) has shown promising results in information retrieval. In essence, DR requires high-quality text representations to support effective search in the representation space. Recent studies have shown that pre-trained autoencoder-based language models with a weak decoder can provide high-quality text representations, boosting the effectiveness and few-shot ability of DR models. However, even a weak autoregressive decoder has the bypass effect on the encoder. More importantly, the discriminative ability of learned representations may be limited since each token is treated equally important in decoding the input texts. To address the above problems, in this paper, we propose a contrastive pre-training approach to learn a discriminative autoencoder with a lightweight multi-layer perception (MLP) decoder. The basic idea is to generate word distributions of input text in a non-autoregressive fashion and pull the word distributions of two masked versions of one text close while pushing away from others. We theoretically show that our contrastive strategy can suppress the common words and highlight the representative words in decoding, leading to discriminative representations. Empirical results show that our method can significantly outperform the state-of-the-art autoencoder-based language models and other pre-trained models for dense retrieval.
Planning-based reinforcement learning has shown strong performance in tasks in discrete and low-dimensional continuous action spaces. However, planning usually brings significant computational overhead for decision-making, and scaling such methods to high-dimensional action spaces remains challenging. To advance efficient planning for high-dimensional continuous control, we propose Trajectory Autoencoding Planner (TAP), which learns low-dimensional latent action codes with a state-conditional VQ-VAE. The decoder of the VQ-VAE thus serves as a novel dynamics model that takes latent actions and current state as input and reconstructs long-horizon trajectories. During inference time, given a starting state, TAP searches over discrete latent actions to find trajectories that have both high probability under the training distribution and high predicted cumulative reward. Empirical evaluation in the offline RL setting demonstrates low decision latency which is indifferent to the growing raw action dimensionality. For Adroit robotic hand manipulation tasks with high-dimensional continuous action space, TAP surpasses existing model-based methods by a large margin and also beats strong model-free actor-critic baselines.
Autoencoders, which consist of an encoder and a decoder, are widely used in machine learning for dimension reduction of high-dimensional data. The encoder embeds the input data manifold into a lower-dimensional latent space, while the decoder represents the inverse map, providing a parametrization of the data manifold by the manifold in latent space. A good regularity and structure of the embedded manifold may substantially simplify further data processing tasks such as cluster analysis or data interpolation. We propose and analyze a novel regularization for learning the encoder component of an autoencoder: a loss functional that prefers isometric, extrinsically flat embeddings and allows to train the encoder on its own. To perform the training it is assumed that for pairs of nearby points on the input manifold their local Riemannian distance and their local Riemannian average can be evaluated. The loss functional is computed via Monte Carlo integration with different sampling strategies for pairs of points on the input manifold. Our main theorem identifies a geometric loss functional of the embedding map as the $\Gamma$-limit of the sampling-dependent loss functionals. Numerical tests, using image data that encodes different explicitly given data manifolds, show that smooth manifold embeddings into latent space are obtained. Due to the promotion of extrinsic flatness, these embeddings are regular enough such that interpolation between not too distant points on the manifold is well approximated by linear interpolation in latent space as one possible postprocessing.
Offline Handwritten Mathematical Expression Recognition (HMER) is a major area in the field of mathematical expression recognition. Offline HMER is often viewed as a much harder problem as compared to online HMER due to a lack of temporal information and variability in writing style. In this paper, we purpose a encoder-decoder model that uses paired adversarial learning. Semantic-invariant features are extracted from handwritten mathematical expression images and their printed mathematical expression counterpart in the encoder. Learning of semantic-invariant features combined with the DenseNet encoder and transformer decoder, helped us to improve the expression rate from previous studies. Evaluated on the CROHME dataset, we have been able to improve latest CROHME 2019 test set results by 4% approx.
The goal of this paper is to interactively refine the automatic segmentation on challenging structures that fall behind human performance, either due to the scarcity of available annotations or the difficulty nature of the problem itself, for example, on segmenting cancer or small organs. Specifically, we propose a novel Transformer-based architecture for Interactive Segmentation (TIS), that treats the refinement task as a procedure for grouping pixels with similar features to those clicks given by the end users. Our proposed architecture is composed of Transformer Decoder variants, which naturally fulfills feature comparison with the attention mechanisms. In contrast to existing approaches, our proposed TIS is not limited to binary segmentations, and allows the user to edit masks for arbitrary number of categories. To validate the proposed approach, we conduct extensive experiments on three challenging datasets and demonstrate superior performance over the existing state-of-the-art methods. The project page is: https://wtliu7.github.io/tis/.
The Solar Dynamics Observatory (SDO), a NASA multi-spectral decade-long mission that has been daily producing terabytes of observational data from the Sun, has been recently used as a use-case to demonstrate the potential of machine learning methodologies and to pave the way for future deep-space mission planning. In particular, the idea of using image-to-image translation to virtually produce extreme ultra-violet channels has been proposed in several recent studies, as a way to both enhance missions with less available channels and to alleviate the challenges due to the low downlink rate in deep space. This paper investigates the potential and the limitations of such a deep learning approach by focusing on the permutation of four channels and an encoder--decoder based architecture, with particular attention to how morphological traits and brightness of the solar surface affect the neural network predictions. In this work we want to answer the question: can synthetic images of the solar corona produced via image-to-image translation be used for scientific studies of the Sun? The analysis highlights that the neural network produces high-quality images over three orders of magnitude in count rate (pixel intensity) and can generally reproduce the covariance across channels within a 1% error. However the model performance drastically diminishes in correspondence of extremely high energetic events like flares, and we argue that the reason is related to the rareness of such events posing a challenge to model training.
In computer vision, multi-label recognition are important tasks with many real-world applications, but classifying previously unseen labels remains a significant challenge. In this paper, we propose a novel algorithm, Aligned Dual moDality ClaSsifier (ADDS), which includes a Dual-Modal decoder (DM-decoder) with alignment between visual and textual features, for open-vocabulary multi-label classification tasks. Then we design a simple and yet effective method called Pyramid-Forwarding to enhance the performance for inputs with high resolutions. Moreover, the Selective Language Supervision is applied to further enhance the model performance. Extensive experiments conducted on several standard benchmarks, NUS-WIDE, ImageNet-1k, ImageNet-21k, and MS-COCO, demonstrate that our approach significantly outperforms previous methods and provides state-of-the-art performance for open-vocabulary multi-label classification, conventional multi-label classification and an extreme case called single-to-multi label classification where models trained on single-label datasets (ImageNet-1k, ImageNet-21k) are tested on multi-label ones (MS-COCO and NUS-WIDE).
This paper presents Z-Code++, a new pre-trained language model optimized for abstractive text summarization. The model extends the state of the art encoder-decoder model using three techniques. First, we use a two-phase pre-training process to improve model's performance on low-resource summarization tasks. The model is first pre-trained using text corpora for language understanding, and then is continually pre-trained on summarization corpora for grounded text generation. Second, we replace self-attention layers in the encoder with disentangled attention layers, where each word is represented using two vectors that encode its content and position, respectively. Third, we use fusion-in-encoder, a simple yet effective method of encoding long sequences in a hierarchical manner. Z-Code++ creates new state of the art on 9 out of 13 text summarization tasks across 5 languages. Our model is parameter-efficient in that it outperforms the 600x larger PaLM-540B on XSum, and the finetuned 200x larger GPT3-175B on SAMSum. In zero-shot and few-shot settings, our model substantially outperforms the competing models.
To generate proper captions for videos, the inference needs to identify relevant concepts and pay attention to the spatial relationships between them as well as to the temporal development in the clip. Our end-to-end encoder-decoder video captioning framework incorporates two transformer-based architectures, an adapted transformer for a single joint spatio-temporal video analysis as well as a self-attention-based decoder for advanced text generation. Furthermore, we introduce an adaptive frame selection scheme to reduce the number of required incoming frames while maintaining the relevant content when training both transformers. Additionally, we estimate semantic concepts relevant for video captioning by aggregating all ground truth captions of each sample. Our approach achieves state-of-the-art results on the MSVD, as well as on the large-scale MSR-VTT and the VATEX benchmark datasets considering multiple Natural Language Generation (NLG) metrics. Additional evaluations on diversity scores highlight the expressiveness and diversity in the structure of our generated captions.