Deep neural networks (DNNs) have shown superior performances on various multimodal learning problems. However, it often requires huge efforts to adapt DNNs to individual multimodal tasks by manually engineering unimodal features and designing multimodal feature fusion strategies. This paper proposes Bilevel Multimodal Neural Architecture Search (BM-NAS) framework, which makes the architecture of multimodal fusion models fully searchable via a bilevel searching scheme. At the upper level, BM-NAS selects the inter/intra-modal feature pairs from the pretrained unimodal backbones. At the lower level, BM-NAS learns the fusion strategy for each feature pair, which is a combination of predefined primitive operations. The primitive operations are elaborately designed and they can be flexibly combined to accommodate various effective feature fusion modules such as multi-head attention (Transformer) and Attention on Attention (AoA). Experimental results on three multimodal tasks demonstrate the effectiveness and efficiency of the proposed BM-NAS framework. BM-NAS achieves competitive performances with much less search time and fewer model parameters in comparison with the existing generalized multimodal NAS methods.
In this paper, we present a regression-based pose recognition method using cascade Transformers. One way to categorize the existing approaches in this domain is to separate them into 1). heatmap-based and 2). regression-based. In general, heatmap-based methods achieve higher accuracy but are subject to various heuristic designs (not end-to-end mostly), whereas regression-based approaches attain relatively lower accuracy but they have less intermediate non-differentiable steps. Here we utilize the encoder-decoder structure in Transformers to perform regression-based person and keypoint detection that is general-purpose and requires less heuristic design compared with the existing approaches. We demonstrate the keypoint hypothesis (query) refinement process across different self-attention layers to reveal the recursive self-attention mechanism in Transformers. In the experiments, we report competitive results for pose recognition when compared with the competing regression-based methods.
Synthetic aperture imaging (SAI) is able to achieve the see through effect by blurring out the off-focus foreground occlusions and reconstructing the in-focus occluded targets from multi-view images. However, very dense occlusions and extreme lighting conditions may bring significant disturbances to the SAI based on conventional frame-based cameras, leading to performance degeneration. To address these problems, we propose a novel SAI system based on the event camera which can produce asynchronous events with extremely low latency and high dynamic range. Thus, it can eliminate the interference of dense occlusions by measuring with almost continuous views, and simultaneously tackle the over/under exposure problems. To reconstruct the occluded targets, we propose a hybrid encoder-decoder network composed of spiking neural networks (SNNs) and convolutional neural networks (CNNs). In the hybrid network, the spatio-temporal information of the collected events is first encoded by SNN layers, and then transformed to the visual image of the occluded targets by a style-transfer CNN decoder. Through experiments, the proposed method shows remarkable performance in dealing with very dense occlusions and extreme lighting conditions, and high quality visual images can be reconstructed using pure event data.
We present a contrasting learning approach with data augmentation techniques to learn document representations in an unsupervised manner. Inspired by recent contrastive self-supervised learning algorithms used for image and NLP pretraining, we hypothesize that high-quality document embedding should be invariant to diverse paraphrases that preserve the semantics of the original document. With different backbones and contrastive learning frameworks, our study reveals the enormous benefits of contrastive augmentation for document representation learning with two additional insights: 1) including data augmentation in a contrastive way can substantially improve the embedding quality in unsupervised document representation learning, and 2) in general, stochastic augmentations generated by simple word-level manipulation work much better than sentence-level and document-level ones. We plug our method into a classifier and compare it with a broad range of baseline methods on six benchmark datasets. Our method can decrease the classification error rate by up to 6.4% over the SOTA approaches on the document classification task, matching or even surpassing fully-supervised methods.
Constrained objects, such as doors and drawers are often complex and share a similar structure in the human environment. A robot needs to interact accurately with constrained objects to safely and successfully complete a task. Learning from Demonstration offers an appropriate path to learn the object structure of the demonstration for unknown objects for unknown tasks. There is work that extracts the kinematic model from motion. However, the gap remains when the robot faces a new object with a similar model but different contexts, e.g. size, appearance, etc. In this paper, we propose a framework that integrates all the information needed to learn a constrained motion from a depth camera into a descriptor of the constrained task. The descriptor consists of object information, grasping point model, constrained model, and reference frame model. By associating constrained learning and reference frame with the constrained object, we demonstrate that the robot can learn the book opening model and parameter of the constraints from demonstration and generalize to novel books.
Graph translation is very promising research direction and has a wide range of potential real-world applications. Graph is a natural structure for representing relationship and interactions, and its translation can encode the intrinsic semantic changes of relationships in different scenarios. However, despite its seemingly wide possibilities, usage of graph translation so far is still quite limited. One important reason is the lack of high-quality paired dataset. For example, we can easily build graphs representing peoples' shared music tastes and those representing co-purchase behavior, but a well paired dataset is much more expensive to obtain. Therefore, in this work, we seek to provide a graph translation model in the semi-supervised scenario. This task is non-trivial, because graph translation involves changing the semantics in the form of link topology and node attributes, which is difficult to capture due to the combinatory nature and inter-dependencies. Furthermore, due to the high order of freedom in graph's composition, it is difficult to assure the generalization ability of trained models. These difficulties impose a tighter requirement for the exploitation of unpaired samples. Addressing them, we propose to construct a dual representation space, where transformation is performed explicitly to model the semantic transitions. Special encoder/decoder structures are designed, and auxiliary mutual information loss is also adopted to enforce the alignment of unpaired/paired examples. We evaluate the proposed method in three different datasets.
Node classification is an important research topic in graph learning. Graph neural networks (GNNs) have achieved state-of-the-art performance of node classification. However, existing GNNs address the problem where node samples for different classes are balanced; while for many real-world scenarios, some classes may have much fewer instances than others. Directly training a GNN classifier in this case would under-represent samples from those minority classes and result in sub-optimal performance. Therefore, it is very important to develop GNNs for imbalanced node classification. However, the work on this is rather limited. Hence, we seek to extend previous imbalanced learning techniques for i.i.d data to the imbalanced node classification task to facilitate GNN classifiers. In particular, we choose to adopt synthetic minority over-sampling algorithms, as they are found to be the most effective and stable. This task is non-trivial, as previous synthetic minority over-sampling algorithms fail to provide relation information for newly synthesized samples, which is vital for learning on graphs. Moreover, node attributes are high-dimensional. Directly over-sampling in the original input domain could generates out-of-domain samples, which may impair the accuracy of the classifier. We propose a novel framework, GraphSMOTE, in which an embedding space is constructed to encode the similarity among the nodes. New samples are synthesize in this space to assure genuineness. In addition, an edge generator is trained simultaneously to model the relation information, and provide it for those new samples. This framework is general and can be easily extended into different variations. The proposed framework is evaluated using three different datasets, and it outperforms all baselines with a large margin.
Synthetic aperture imaging (SAI) is able to achieve the see through effect by blurring out the off-focus foreground occlusions and reconstructing the in-focus occluded targets from multi-view images. However, very dense occlusions and extreme lighting conditions may bring significant disturbances to SAI based on conventional frame-based cameras, leading to performance degeneration. To address these problems, we propose a novel SAI system based on the event camera which can produce asynchronous events with extremely low latency and high dynamic range. Thus, it can eliminate the interference of dense occlusions by measuring with almost continuous views, and simultaneously tackle the over/under exposure problems. To reconstruct the occluded targets, we propose a hybrid encoder-decoder network composed of spiking neural networks (SNNs) and convolutional neural networks (CNNs). In the hybrid network, the spatio-temporal information of the collected events is first encoded by SNN layers, and then transformed to the visual image of the occluded targets by a style-transfer CNN decoder. Through experiments, the proposed method shows remarkable performance in dealing with very dense occlusions and extreme lighting conditions, and high quality visual images can be reconstructed using pure event data.
Many manipulation tasks require robots to interact with unknown environments. In such applications, the ability to adapt the impedance according to different task phases and environment constraints is crucial for safety and performance. Although many approaches based on deep reinforcement learning (RL) and learning from demonstration (LfD) have been proposed to obtain variable impedance skills on contact-rich manipulation tasks, these skills are typically task-specific and could be sensitive to changes in task settings. This paper proposes an inverse reinforcement learning (IRL) based approach to recover both the variable impedance policy and reward function from expert demonstrations. We explore different action space of the reward functions to achieve a more general representation of expert variable impedance skills. Experiments on two variable impedance tasks (Peg-in-Hole and Cup-on-Plate) were conducted in both simulations and on a real FANUC LR Mate 200iD/7L industrial robot. The comparison results with behavior cloning and force-based IRL proved that the learned reward function in the gain action space has better transferability than in the force space. Experiment videos are available at https://msc.berkeley.edu/research/impedance-irl.html.