Visual dialog is a task of answering a sequence of questions grounded in an image using the previous dialog history as context. In this paper, we study how to address two fundamental challenges for this task: (1) reasoning over underlying semantic structures among dialog rounds and (2) identifying several appropriate answers to the given question. To address these challenges, we propose a Sparse Graph Learning (SGL) method to formulate visual dialog as a graph structure learning task. SGL infers inherently sparse dialog structures by incorporating binary and score edges and leveraging a new structural loss function. Next, we introduce a Knowledge Transfer (KT) method that extracts the answer predictions from the teacher model and uses them as pseudo labels. We propose KT to remedy the shortcomings of single ground-truth labels, which severely limit the ability of a model to obtain multiple reasonable answers. As a result, our proposed model significantly improves reasoning capability compared to baseline methods and outperforms the state-of-the-art approaches on the VisDial v1.0 dataset. The source code is available at https://github.com/gicheonkang/SGLKT-VisDial.
We present a new form of ensemble method–Devil’s Advocate, which uses a deliberately dissenting model to force other submodels within the ensemble to better collaborate. Our method consists of two different training settings: one follows the conventional training process (Norm), and the other is trained by artificially generated labels (DevAdv). After training the models, Norm models are fine-tuned through an additional loss function, which uses the DevAdv model as a constraint. In making a final decision, the proposed ensemble model sums the scores of Norm models and then subtracts the score of the DevAdv model. The DevAdv model improves the overall performance of the other models within the ensemble. In addition to our ensemble framework being based on psychological background, it also shows comparable or improved performance on 5 text classification tasks when compared to conventional ensemble methods.
Video Question Answering is a task which requires an AI agent to answer questions grounded in video. This task entails three key challenges: (1) understand the intention of various questions, (2) capturing various elements of the input video (e.g., object, action, causality), and (3) cross-modal grounding between language and vision information. We propose Motion-Appearance Synergistic Networks (MASN), which embed two cross-modal features grounded on motion and appearance information and selectively utilize them depending on the question’s intentions. MASN consists of a motion module, an appearance module, and a motion-appearance fusion module. The motion module computes the action-oriented cross-modal joint representations, while the appearance module focuses on the appearance aspect of the input video. Finally, the motion-appearance fusion module takes each output of the motion module and the appearance module as input, and performs question-guided fusion. As a result, MASN achieves new state-of-the-art performance on the TGIF-QA and MSVD-QA datasets. We also conduct qualitative analysis by visualizing the inference results of MASN.
Scene graph is a graph representation that explicitly represents high-level semantic knowledge of an image such as objects, attributes of objects and relationships between objects. Various tasks have been proposed for the scene graph, but the problem is that they have a limited vocabulary and biased information due to their own hypothesis. Therefore, results of each task are not generalizable and difficult to be applied to other down-stream tasks. In this paper, we propose Entity Synset Alignment(ESA), which is a method to create a general scene graph by aligning various semantic knowledge efficiently to solve this bias problem. The ESA uses a large-scale lexical database, WordNet and Intersection of Union (IoU) to align the object labels in multiple scene graphs/semantic knowledge. In experiment, the integrated scene graph is applied to the image-caption retrieval task as a down-stream task. We confirm that integrating multiple scene graphs helps to get better representations of images.
In this work, we propose a goal-driven collaborative task that combines language, perception, and action. Specifically, we develop a Collaborative image-Drawing game between two agents, called CoDraw. Our game is grounded in a virtual world that contains movable clip art objects. The game involves two players: a Teller and a Drawer. The Teller sees an abstract scene containing multiple clip art pieces in a semantically meaningful configuration, while the Drawer tries to reconstruct the scene on an empty canvas using available clip art pieces. The two players communicate with each other using natural language. We collect the CoDraw dataset of ~10K dialogs consisting of ~138K messages exchanged between human players. We define protocols and metrics to evaluate learned agents in this testbed, highlighting the need for a novel “crosstalk” evaluation condition which pairs agents trained independently on disjoint subsets of the training data. We present models for our task and benchmark them using both fully automated evaluation and by having them play the game live with humans.
Visual dialog (VisDial) is a task which requires a dialog agent to answer a series of questions grounded in an image. Unlike in visual question answering (VQA), the series of questions should be able to capture a temporal context from a dialog history and utilizes visually-grounded information. Visual reference resolution is a problem that addresses these challenges, requiring the agent to resolve ambiguous references in a given question and to find the references in a given image. In this paper, we propose Dual Attention Networks (DAN) for visual reference resolution in VisDial. DAN consists of two kinds of attention modules, REFER and FIND. Specifically, REFER module learns latent relationships between a given question and a dialog history by employing a multi-head attention mechanism. FIND module takes image features and reference-aware representations (i.e., the output of REFER module) as input, and performs visual grounding via bottom-up attention mechanism. We qualitatively and quantitatively evaluate our model on VisDial v1.0 and v0.9 datasets, showing that DAN outperforms the previous state-of-the-art model by a significant margin.
A statistical method for compound noun decomposition is presented. Previous studies on this problem showed some statistical information are helpful. But applying statistical information was not so systemic that performance depends heavily on the algorithm and some algorithms usually have many separated steps. In our work statistical information is collected from manually decomposed compound noun corpus to build a Markov model for composition. Two Markov chains representing statistical information are assumed independent: one for the sequence of participants' lengths and another for the sequence of participants ' features. Besides Markov assumptions, least participants preference assumption also is used. These two assumptions enable the decomposition algorithm to be a kind of conditional dynamic programming so that efficient and systemic computation can be performed. When applied to test data of size 5027, we obtained a precision of 98.4%.