Large language models (LLMs) are revolutionizing user interactions in online map applications by enabling conversational interfaces with intelligent map agents. However, inherent ambiguities and nuances in human communication often lead to incomplete or unclear user instructions. While generating clarification questions can mitigate this issue, existing methods—which rely on prompt-based LLM assessments or supervised fine-tuning (SFT) of LLMs with limited annotated data—struggle to reliably determine when clarification is necessary, resulting in cold-start issues and reduced adaptability. To address this gap, we propose an uncertainty-aware dynamic planning framework for intent disambiguation in interactive agent systems, exemplified by Baidu Maps. Our framework leverages LLMs to dynamically generate agent action sequences while estimating query uncertainty. When uncertainty thresholds are exceeded, targeted clarification questions are triggered. Besides, to efficiently navigate the vast uncertainty space, we introduce a novel beam search-based pruning strategy. Crucially, the framework supports context-aware task planning without any additional model training. We evaluate our approach on Baidu Maps and a public robotic task, focusing on uncertainty management and adaptability. Experiments demonstrate that the framework achieves high precision across diverse LLMs without requiring an additional training phase, highlighting its versatility and robustness. Ablation studies confirm that the framework can significantly improve user intent recognition and task success rates.

Deep learning models commonly benefit from data augmentation techniques to diversify the set of training images. When working with satellite imagery, it is common for practitioners to apply a limited set of transformations developed for natural images (e.g., flip and rotate) to expand the training set without overly modifying the satellite images. There are many techniques for natural image data augmentation, but given the differences between the two domains, it is not clear whether data augmentation methods developed for natural images are well suited for satellite imagery. This paper presents an extensive experimental study on three classification and three regression tasks over four satellite image datasets. We compare common computer vision data augmentation techniques and propose three novel satellite-specific data augmentation strategies. Across tasks and datasets, we find that geometric transformations are beneficial for satellite imagery while color transformations generally are not. Additionally, our novel Sat-SlideMix, Sat-CutMix, and Sat-Trivial methods all exhibit strong performance across all tasks and datasets.

We propose Long-LRM, a feed-forward 3D Gaussian reconstruction model for instant, high-resolution, 360° wide-coverage, scene-level reconstruction. Specifically, it takes in 32 input images at a resolution of 960x540 and produces the Gaussian reconstruction in just 1 second on a single A100 GPU. To handle the long sequence of 250K tokens brought by the large input size, Long-LRM features a mixture of the recent Mamba2 blocks and the classical transformer blocks, enhanced by a light-weight token merging module and Gaussian pruning steps that balance between quality and efficiency. We evaluate Long-LRM on the large-scale DL3DV benchmark and Tanks&Temples, demonstrating reconstruction quality comparable to the optimization-based methods while achieving an 800x speedup w.r.t. the optimization-based approaches and an input size at least 60x larger than the previous feed-forward approaches. We conduct extensive ablation studies on our model design choices for both rendering quality and computation efficiency. We also explore Long-LRM's compatibility with other Gaussian variants such as 2D GS, which enhances Long-LRM's ability in geometry reconstruction. Project page: https://arthurhero.github.io/projects/llrm

Recent point-based object completion methods have demonstrated the ability to accurately recover the missing geometry of partially observed objects. However, these approaches are not well-suited for completing objects within a scene, as they do not consider known scene constraints (e.g., other observed surfaces) in their completions and further expect the partial input to be in a canonical coordinate system, which does not hold for objects within scenes. While instance scene completion methods have been proposed for completing objects within a scene, they lag behind point-based object completion methods in terms of object completion quality and still do not consider known scene constraints during completion. To overcome these limitations, we propose a point cloud-based instance completion model that can robustly complete objects at arbitrary scales and pose in the scene. To enable reasoning at the scene level, we introduce a sparse set of scene constraints represented as point clouds and integrate them into our completion model via a cross-attention mechanism. To evaluate the instance scene completion task on indoor scenes, we further build a new dataset called ScanWCF, which contains labeled partial scans as well as aligned ground truth scene completions that are watertight and collision-free. Through several experiments, we demonstrate that our method achieves improved fidelity to partial scans, higher completion quality, and greater plausibility over existing state-of-the-art methods.

This paper introduces the Convex Potential Mirror Langevin Algorithm (CPMLA), a novel method to improve sampling efficiency for Energy-Based Models (EBMs). CPMLA uses mirror Langevin dynamics with a convex potential flow as a dynamic mirror map for EBM sampling. This dynamic mirror map enables targeted geometric exploration on the data manifold, accelerating convergence to the target distribution. Theoretical analysis proves that CPMLA achieves exponential convergence with vanishing bias under relaxed log-concave conditions, supporting its efficiency in adapting to complex data distributions. Experiments on benchmarks like CIFAR-10, SVHN, and CelebA demonstrate CPMLA's improved sampling quality and inference efficiency over existing techniques.

In order to gain insights about the decision-making of different visual recognition backbones, we propose two methodologies, sub-explanation counting and cross-testing, that systematically applies deep explanation algorithms on a dataset-wide basis, and compares the statistics generated from the amount and nature of the explanations. These methodologies reveal the difference among networks in terms of two properties called compositionality and disjunetivism. Transformers and ConvNeXt are found to be more compositional, in the sense that they jointly consider multiple parts of the image in building their decisions, whereas traditional CNNs and distilled transformers are less compositional and more disjunctive, which means that they use multiple diverse but smaller set of parts to achieve a confident prediction. Through further experiments, we pin-pointed the choice of normalization to be especially important in the compositionality of a model, in that batch normalization leads to less compositionality while group and layer normalization lead to more. Finally, we also analyze the features shared by different backbones and plot a land-scape of different models based on their feature-use similarity.

Despite extensive research on training generative adversarial networks (GANs) with limited training data, learning to generate images from long-tailed training distributions remains fairly unexplored. In the presence of imbalanced multiclass training data, GANs tend to favor classes with more samples, leading to the generation of low quality and less diverse samples in tail classes. In this study, we aim to improve the training of class-conditional GANs with long-tailed data. We propose a straightforward yet effective method for knowledge sharing, allowing tail classes to borrow from the rich information from classes with more abundant training data. More concretely, we propose modifications to existing class-conditional GAN architectures to ensure that the lower-resolution layers of the generator are trained entirely unconditionally while reserving class-conditional generation for the higher-resolution layers. Experiments on several long-tail benchmarks and GAN architectures demonstrate a significant improvement over existing methods in both the diversity and fidelity of the generated images. The code is available at https:github.com/khorrams/utlo.

Modeling object dynamics with a neural network is an important problem with numerous applications. Most recent work has been based on graph neural networks. However, physics happens in 3D space, where geometric information potentially plays an important role in modeling physical phenomena. In this work, we propose a novel U-net architecture based on continuous point convolution which naturally embeds information from 3D coordinates and allows for multi-scale feature representations with established downsampling and upsampling procedures. Bottleneck layers in the downsampled point clouds lead to better long-range interaction modeling. Besides, the flexibility of point convolutions allows our approach to generalize to sparsely sampled points from mesh vertices and dynamically generate features on important interaction points on mesh faces. Experimental results demonstrate that our approach significantly improves the state-of-the-art, especially in scenarios that require accurate gravity or collision reasoning.

Robots need to have a memory of previously observed, but currently occluded objects to work reliably in realistic environments. We investigate the problem of encoding object-oriented memory into a multi-object manipulation reasoning and planning framework. We propose DOOM and LOOM, which leverage transformer relational dynamics to encode the history of trajectories given partial-view point clouds and an object discovery and tracking engine. Our approaches can perform multiple challenging tasks including reasoning with occluded objects, novel objects appearance, and object reappearance. Throughout our extensive simulation and real-world experiments, we find that our approaches perform well in terms of different numbers of objects and different numbers of distractor actions. Furthermore, we show our approaches outperform an implicit memory baseline.

Visual control policies can encounter significant performance degradation when visual conditions like lighting or camera position differ from those seen during training – often exhibiting sharp declines in capability even for minor differences. In this work, we examine robustness to a suite of these types of visual changes for RGB-D and point cloud based visual control policies. To perform these experiments on both model-free and model-based reinforcement learners, we introduce a novel Point Cloud World Model (PCWM) and point cloud based control policies. Our experiments show that policies that explicitly encode point clouds are significantly more robust than their RGB-D counterparts. Further, we find our proposed PCWM significantly outperforms prior works in terms of sample efficiency during training. Taken together, these results suggest reasoning about the 3D scene through point clouds can improve performance, reduce learning time, and increase robustness for robotic learners. Code: https://github.com/pvskand/pcwm

Deep learning has brought transformative advancements to object segmentation, especially in marine robotics contexts such as waste management and subaquatic infrastructure oversight. However, a central challenge persists: calibrating the prediction confidence of the model to ensure robust and reliable outcomes, especially within the demanding underwater environment. Existing solutions for estimating uncertainty are often computationally intensive and have largely centered around Bayesian neural networks or ensemble methods. In this paper, we present a Conditional Variational Autoencoder-based framework (CVAE-SM), which is capable of generating diverse latent codes for improved uncertainty quantification in image segmentation. Our method, enhanced by a style modulator, merges content features, and latent codes more effectively, leading to refined prediction of uncertainty levels. We further introduce a dataset of perturbed underwater images to benchmark uncertainty quantification in this domain. The proposed model not only surpasses peers in segmentation metrics but also matches ensemble models in uncertainty predictions, all while being 2.5 times faster.

Particle-based Bayesian deep learning often requires a similarity metric to compare two networks. However, naive similarity metrics lack permutation invariance and are inappropriate for comparing networks. Centered Kernel Alignment (CKA) on feature kernels has been proposed to compare deep networks but has not been used as an optimization objective in Bayesian deep learning. In this paper, we explore the use of CKA in Bayesian deep learning to generate diverse ensembles and hypernetworks that output a network posterior. Noting that CKA projects kernels onto a unit hypersphere and that directly optimizing the CKA objective leads to diminishing gradients when two networks are very similar. We propose adopting the approach of hyperspherical energy (HE) on top of CKA kernels to address this drawback and improve training stability. Additionally, by leveraging CKA-based feature kernels, we derive feature repulsive terms applied to synthetically generated outlier examples. Experiments on both diverse ensembles and hypernetworks show that our approach significantly outperforms baselines in terms of uncertainty quantification in both synthetic and realistic outlier detection tasks.

Real world images often have highly imbalanced content density. Some areas are very uniform, e.g., large patches of blue sky, while other areas are scattered with many small objects. Yet, the commonly used successive grid downsampling strategy in convolutional deep networks treats all areas equally. Hence, small objects are represented in very few spatial locations, leading to worse results in tasks such as segmentation. Intuitively, retaining more pixels representing small objects during downsampling helps to preserve important information. To achieve this, we propose AutoFocusFormer (AFF), a local-attention transformer image recognition backbone, which performs adaptive down-sampling by learning to retain the most important pixels for the task. Since adaptive downsampling generates a set of pixels irregularly distributed on the image plane, we abandon the classic grid structure. Instead, we develop a novel point-based local attention block, facilitated by a balanced clustering module and a learnable neighborhood merging module, which yields representations for our point-based versions of state-of-the-art segmentation heads. Experiments show that our AutoFocusFormer (AFF) improves significantly over baseline models of similar sizes.

We introduce PointConvFormer, a novel building block for point cloud based deep network architectures. Inspired by generalization theory, PointConvFormer combines ideas from point convolution, where filter weights are only based on relative position, and Transformers which utilize feature-based attention. In PointConvFormer, attention computed from feature difference between points in the neighborhood is used to modify the convolutional weights at each point. Hence, we preserved the invariances from point convolution, whereas attention helps to select relevant points in the neighborhood for convolution. PointConvFormer is suitable for multiple tasks that require details at the point level, such as segmentation and scene flow estimation tasks. We experiment on both tasks with multiple datasets including ScanNet, SemanticKitti, FlyingThings3D and KITTI. Our results show that PointConvFormer offers a better accuracy-speed tradeoff than classic convolutions, regular transformers, and voxelized sparse convolution approaches. Visualizations show that PointConvFormer performs similarly to convolution on flat areas, whereas the neighborhood selection effect is stronger on object boundaries, showing that it has got the best of both worlds. The code will be available.

Robots performing mobile manipulation in unstructured environments must identify grasp affordances quickly and with robustness to perception noise. Yet in domains such as underwater manipulation, where perception noise is severe, computation is constrained, and the environment is dynamic, existing techniques fail. They are too computationally demanding, or too sensitive to noise to allow for closed loop grasping or dynamic replanning, or do not consider 6-DOF grasps. We present a novel grasp synthesis network, TSGrasp, that uses spatio-temporal sparse convolution to process a streaming point cloud in real time. The network generates 6-DOF grasps at greater speed and with less memory than Contact GraspNet, a state-of-the-art algorithm based on Point-Net++. By considering information from multiple successive frames of depth video, TSGrasp boosts robustness to noise or temporary self-occlusion and allows more grasps to be rapidly identified. Our grasp synthesis system was successfully demonstrated in an underwater environment with a Blueprint Labs Bravo robotic arm.

Shape completion aims to recover the full 3D geometry of an object from a partial observation. This problem is inherently multi-modal since there can be many ways to plausibly complete the missing regions of a shape. Such diversity would be indicative of the underlying uncertainty of the shape and could be preferable for downstream tasks such as planning. In this paper, we propose a novel conditional generative adversarial network that can produce many diverse plausible completions of a partially observed point cloud. To enable our network to produce multiple completions for the same partial input, we introduce stochasticity into our network via style modulation. By extracting style codes from complete shapes during training, and learning a distribution over them, our style codes can explicitly carry shape category information leading to better completions. We further introduce diversity penalties and discriminators at multiple scales to prevent conditional mode collapse and to train without the need for multiple ground truth completions for each partial input. Evaluations across several synthetic and real datasets demonstrate that our method achieves significant improvements in respecting the partial observations while obtaining greater diversity in completions.

Recently, there is significant interest in performing convolution over irregularly sampled point clouds. Point clouds are very different from raster images, in that one cannot have a regular sampling grid on point clouds, which makes robustness under irregular neighborhoods an important issue. Especially, the k-nearest neighbor (kNN) neighborhood presents challenges for generalization because the location of the neighbors can be very different between training and testing times. In order to improve the robustness to different neighborhood samplings, this paper proposes a novel viewpoint-invariant coordinate transform as the input to the weight-generating function for point convolution, in addition to the regular 3D coordinates. This allows us to feed the network with non-invariant, scale-invariant and scale+rotation-invariant coordinates simultaneously, so that the network can learn which to include in the convolution function automatically. Empirically, we demonstrate that this effectively improves the performance of point cloud convolutions on the SemanticKITTI and ScanNet datasets, as well as the robustness to significant test-time downsampling, which can substantially change the distance of neighbors in a kNN neighborhood. Experimentally, among pure point-based approaches, we achieve comparable semantic segmentation performance with a comparable point-based convolution framework KPConv on SemanticKITTI and ScanNet, yet is significantly more efficient by virtue of using a kNN neighborhood instead of an ϵ-ball.

CounterFactual (CF) visual explanations try to find images similar to the query image that change the decision of a vision system to a specified outcome. Existing methods either require inference-time optimization or joint training with a generative adversarial model which makes them time-consuming and difficult to use in practice. We propose a novel approach, Cycle-Consistent Counterfactuals by Latent Transformations (C3LT), which learns a latent transformation that automatically generates visual CFs by steering in the latent space of generative models. Our method uses cycle consistency between the query and CF latent representations which helps our training to find better solutions. C3LT can be easily plugged into any state-of-the-art pretrained generative network. This enables our method to generate high-quality and interpretable CF images at high resolution such as those in ImageNet. In addition to several established metrics for evaluating CF explanations, we introduce a novel metric tailored to assess the quality of the generated CF examples and validate the effectiveness of our method on an extensive set of experiments.

In multi-object tracking, the tracker maintains in its memory the appearance and motion information for each object in the scene. This memory is utilized for finding matches between tracks and detections, and is updated based on the matching. Many approaches model each target in isolation and lack the ability to use all the targets in the scene to jointly update the memory. This can be problematic when there are similarly looking objects in the scene. In this paper, we solve the problem of simultaneously considering all tracks during memory updating, with only a small spatial overhead, via a novel multi-track pooling module. We additionally propose a training strategy adapted to multi-track pooling which generates hard tracking episodes online. We show that the combination of these innovations results in a strong discriminative appearance model under the bilinear LSTM tracking framework, enabling the use of greedy data association to achieve online tracking performance. Our experiments demonstrate real-time, state-of-the-art online tracking performance on public multi-object tracking (MOT) datasets. The code and trained models are available at https://github.com/chkim403/blstm-mtp.

We present MetaUVFS as the first Unsupervised Meta-learning algorithm for Video Few-Shot action recognition. MetaUVFS leverages over 550K unlabeled videos to train a two-stream 2D and 3D CNN architecture via contrastive learning to capture the appearance-specific spatial and action-specific spatio-temporal video features respectively. MetaUVFS comprises a novel Action-Appearance Aligned Meta-adaptation (A3M) module that learns to focus on the action-oriented video features in relation to the appearance features via explicit few-shot episodic meta-learning over unsupervised hard-mined episodes. Our action-appearance alignment and explicit few-shot learner conditions the unsupervised training to mimic the downstream few-shot task, enabling MetaUVFS to significantly outperform all state-of-the-art unsupervised methods on few-shot benchmarks. Moreover, unlike previous few-shot action recognition methods that are supervised, MetaUVFS needs neither base-class labels nor a supervised pretrained backbone. Thus, we need to train MetaUVFS just once to perform competitively or sometimes even outperform state-of-the-art supervised methods on popular HMDB51, UCF101, and Kinetics100 few-shot datasets.

Recently, particle-based variational inference (ParVI) methods have gained interest because they can avoid arbitrary parametric assumptions that are common in variational inference. However, many ParVI approaches do not allow arbitrary sampling from the posterior, and the few that do allow such sampling suffer from suboptimality. This work proposes a new method for learning to approximately sample from the posterior distribution. We construct a neural sampler that is trained with the functional gradient of the KL-divergence between the empirical sampling distribution and the target distribution, assuming the gradient resides within a reproducing kernel Hilbert space. Our generative ParVI (GPVI) approach maintains the asymptotic performance of ParVI methods while offering the flexibility of a generative sampler. Through carefully constructed experiments, we show that GPVI outperforms previous generative ParVI methods such as amortized SVGD, and is competitive with ParVI as well as gold-standard approaches like Hamiltonian Monte Carlo for fitting both exactly known and intractable target distributions.

Attention maps are popular tools for explaining the decisions of convolutional neural networks (CNNs) for image classification. Typically, for each image of interest, a single attention map is produced, which assigns weights to pixels based on their importance to the classification. We argue that a single attention map provides an incomplete understanding since there are often many other maps that explain a classification equally well. In this paper, we propose to utilize a beam search algorithm to systematically search for multiple explanations for each image. Results show that there are indeed multiple relatively localized explanations for many images. However, naively showing multiple explanations to users can be overwhelming and does not reveal their common and distinct structures. We introduce structured attention graphs (SAGs), which compactly represent sets of attention maps for an image by visualizing how different combinations of image regions impact the confidence of a classifier. An approach to computing a compact and representative SAG for visualization is proposed via diverse sampling. We conduct a user study comparing the use of SAGs to traditional attention maps for answering comparative counterfactual questions about image classifications. Our results show that the users are significantly more accurate when presented with SAGs compared to standard attention map baselines.

Due to the high representational efficiency, superpixel largely reduces the number of image primitives for subsequent processing. However, superpixel is scarcely utilized in recent methods since its irregular shape is intractable for standard convolutional layer. In this paper, we propose an end-to-end trainable superpixel convolutional network, named SPNet, to learn high-level representation on image superpixel primitives. We start by treating irregular superpixel lattices as a 2D point cloud, where the low-level features inside one superpixel are aggregated to one feature vector. We replace the standard convolutional layer with the PointConv layer to handle the irregular and unordered point cloud. Besides, we propose grid based downsampling strategies to output uniform 2D sampling result. The resulting network largely utilizes the efficiency of superpixel and provides a novel view for image recognition task. Experiments on image recognition task show promising results compared with prominent image classification methods. The visualization of class activation mapping shows great accuracy at object localization and boundary segmentation.

Understanding and interpreting the decisions made by deep learning models is valuable in many domains. In computer vision, computing heatmaps from a deep network is a popular approach for visualizing and understanding deep networks. However, heatmaps that do not correlate with the network may mislead human, hence the performance of heatmaps in providing a faithful explanation to the underlying deep network is crucial. In this paper, we propose I-GOS, which optimizes for a heatmap so that the classification scores on the masked image would maximally decrease. The main novelty of the approach is to compute descent directions based on the integrated gradients instead of the normal gradient, which avoids local optima and speeds up convergence. Compared with previous approaches, our method can flexibly compute heatmaps at any resolution for different user needs. Extensive experiments on several benchmark datasets show that the heatmaps produced by our approach are more correlated with the decision of the underlying deep network, in comparison with other state-of-the-art approaches.

Efficient exploration remains a challenging problem in reinforcement learning, especially for those tasks where rewards from environments are sparse. In this work, we introduce an exploration approach based on a novel implicit generative modeling algorithm to estimate a Bayesian uncertainty of the agent’s belief of the environment dynamics. Each random draw from our generative model is a neural network that instantiates the dynamic function, hence multiple draws would approximate the posterior, and the variance in the predictions based on this posterior is used as an intrinsic reward for exploration. We design a training algorithm for our generative model based on the amortized Stein Variational Gradient Descent. In experiments, we demonstrate the effectiveness of this exploration algorithm in both pure exploration tasks and a downstream task, comparing with state-of-the-art intrinsic reward-based exploration approaches, including two recent approaches based on an ensemble of dynamic models. In challenging exploration tasks, our implicit generative model consistently outperforms competing approaches regarding data efficiency in exploration.

Instance segmentation, which seeks to obtain both class and instance labels for each pixel in the input image, is a challenging task in computer vision. State-of- the-art algorithms often employ a search-based strategy, which first divides the output image with a regular grid and generate proposals at each grid cell, then the proposals are classified and boundaries refined. In this paper, we propose a novel algorithm that directly utilizes a fully convolutional network (FCN) to predict instance labels. Specifically, we propose a variational relaxation of instance segmentation as minimizing an optimization functional for a piecewise-constant segmentation problem, which can be used to train an FCN end-to-end. It extends the classical Mumford-Shah variational segmentation algorithm to be able to handle the permutation-invariant ground truth in instance segmentation. Experiments on PASCAL VOC 2012 and the MSCOCO 2017 dataset show that the proposed approach efficiently tackles the instance segmentation task.

Unlike images which are represented in regular dense grids, 3D point clouds are irregular and unordered, hence applying convolution on them can be difficult. In this paper, we extend the dynamic filter to a new convolution operation, named PointConv. PointConv can be applied on point clouds to build deep convolutional networks. We treat convolution kernels as nonlinear functions of the local coordinates of 3D points comprised of weight and density functions. With respect to a given point, the weight functions are learned with multi-layer perceptron networks and the density functions through kernel density estimation. A novel reformulation is proposed for efficiently computing the weight functions, which allowed us to dramatically scale up the network and significantly improve its performance. The learned convolution kernel can be used to compute translation-invariant and permutation-invariant convolution on any point set in the 3D space. Besides, PointConv can also be used as deconvolution operators to propagate features from a subsampled point cloud back to its original resolution. Experiments on ModelNet40, ShapeNet, and ScanNet show that deep convolutional neural networks built on PointConv are able to achieve state-of-the-art on challenging semantic segmentation benchmarks on 3D point clouds. Besides, our experiments converting CIFAR-10 into a point cloud showed that networks built on PointConv can match the performance of convolutional networks in 2D images of a similar structure.

Heatmap regression with a deep network has become one of the mainstream approaches to localize facial landmarks. However, the loss function for heatmap regression is rarely studied. In this paper, we analyze the ideal loss function properties for heatmap regression in face alignment problems. Then we propose a novel loss function, named Adaptive Wing loss, that is able to adapt its shape to different types of ground truth heatmap pixels. This adaptability penalizes loss more on foreground pixels while less on background pixels. To address the imbalance between foreground and background pixels, we also propose Weighted Loss Map, which assigns high weights on foreground and difficult background pixels to help training process focus more on pixels that are crucial to landmark localization. To further improve face alignment accuracy, we introduce boundary prediction and CoordConv with boundary coordinates. Extensive experiments on different benchmarks, including COFW, 300W and WFLW, show our approach outperforms the state-of-the-art by a significant margin on various evaluation metrics. Besides, the Adaptive Wing loss also helps other heatmap regression tasks.

We introduce HyperGAN, a generative model that learns to generate all the parameters of a deep neural network. HyperGAN first transforms low dimensional noise into a latent space, which can be sampled from to obtain diverse, performant sets of parameters for a target architecture. We utilize an architecture that bears resemblance to generative adversarial networks, but we evaluate the likelihood of generated samples with a classification loss. This is equivalent to minimizing the KL-divergence between the distribution of generated parameters, and the unknown true parameter distribution. We apply HyperGAN to classification, showing that HyperGAN can learn to generate parameters which solve the MNIST and CIFAR-10 datasets with competitive performance to fully supervised learning, while also generating a rich distribution of effective parameters. We also show that HyperGAN can also provide better uncertainty estimates than standard ensembles. This is evidenced by the ability of HyperGAN-generated ensembles to detect out of distribution data as well as adversarial examples.

In this work we address the challenge of predicting the missing dimension (elevation angle) from 2D underwater sonar images. The high noise levels in these images, from phenomena such as non-diffuse reflections, frequently limits the usefulness of physical models. We thus propose the utilization of Convolutional Neural Networks (CNNs) as a powerful method to extract meaningful information without being misled by noisy data. We also introduce a self-supervised method that uses the physics of the sonar sensor to train the network on real data without ground-truth elevation maps. Our method can produce accurate elevation angle estimates given only a single image. Finally, we demonstrate that our method produces more accurate 3D reconstructions than competing methods, both in simulation and on real data.

Segmentation algorithms are prone to make topological errors on fine-scale struc- tures, e.g., broken connections. We propose a novel method that learns to segment with correct topology. In particular, we design a continuous-valued loss function that enforces a segmentation to have the same topology as the ground truth, i.e.,having the same Betti number. The proposed topology-preserving loss function is differentiable and can be incorporated into end-to-end training of a deep neural network. Our method achieves much better performance on the Betti number error, which directly accounts for the topological correctness. It also performs superior on other topology-relevant metrics, e.g., the Adjusted Rand Index and the Variation of Information, without sacrificing per-pixel accuracy. We illustrate the effectiveness of the proposed method on a broad spectrum of natural and biomedical datasets.

Using deep learning, this paper addresses the problem of joint object boundary detection and boundary motion estimation in videos, which we named boundary flow estimation. Boundary flow is an important mid-level visual cue as boundaries characterize objects' spatial extents, and the flow indicates objects' motions and interactions. Yet, most prior work on motion estimation has focused on dense object motion or feature points that may not necessarily reside on boundaries. For boundary flow estimation, we specify a new fully convolutional Siamese network (FCSN) that jointly estimates object-level boundaries in two consecutive frames. Boundary correspondences in the two frames are predicted by the same FCSN with a new, unconventional deconvolution approach. Finally, the boundary flow estimate is improved with an edgelet-based filtering. Evaluation is conducted on three tasks: boundary detection in videos, boundary flow estimation, and optical flow estimation. On boundary detection, we achieve the state-of-the-art performance on the benchmark VSB100 dataset. On boundary flow estimation, we present the first results on the Sintel training dataset. For optical flow estimation, we run the recent approach CPM-Flow but on the augmented input with our boundary-flow matches, and achieve significant performance improvement on the Sintel benchmark.

In this work we develop a novel framework that enables the real-time 3D reconstruction of underwater environments using features from 2D sonar images. Due to noisy and low-resolution imagery as compared with standard cameras, automatic feature extractors for sonar images are not reliable in many scenarios. Thus, a human often needs to hand-select features in sonar imagery for environment reconstructions. Given the high data capture rates of standard imaging sonars (on the order of 20Hz), hand-annotating the features in every frame cannot be done in real-time. To address this we use a Convolutional Neural Network (CNN) that analyzes incoming imagery in real-time and proposes only a small subset of high-quality frames to the user for feature annotation. We demonstrate that our approach provides real-time reconstruction capability without loss in classification performance on datasets captured onboard our underwater vehicle while operating in a variety of environments.

In this article, we present an interactive visual information retrieval and recommendation system, called VisIRR, for large-scale document discovery. VisIRR effectively combines the paradigms of (1) a passive pull through query processes for retrieval and (2) an active push that recommends items of potential interest to users based on their preferences. Equipped with an efficient dynamic query interface against a large-scale corpus, VisIRR organizes the retrieved documents into high-level topics and visualizes them in a 2D space, representing the relationships among the topics along with their keyword summary. In addition, based on interactive personalized preference feedback with regard to documents, VisIRR provides document recommendations from the entire corpus, which are beyond the retrieved sets. Such recommended documents are visualized in the same space as the retrieved documents, so that users can seamlessly analyze both existing and newly recommended ones. This article presents novel computational methods, which make these integrated representations and fast interactions possible for a large-scale document corpus. We illustrate how the system works by providing detailed usage scenarios. Additionally, we present preliminary user study results for evaluating the effectiveness of the system.

Deep learning has greatly improved visual recognition in recent years. However, recent research has shown that there exist many adversarial examples that can negatively impact the performance of such an architecture. This paper focuses on detecting those adversarial examples by analyzing whether they come from the same distribution as the normal examples. Instead of directly training a deep neural network to detect adversarials, a much simpler approach was proposed based on statistics on outputs from convolutional layers. A cascade classifier was designed to efficiently detect adversarials. Furthermore, trained from one particular adversarial generating mechanism, the resulting classifier can successfully detect adversarials from a completely different mechanism as well. The resulting classifier is non-subdifferentiable, hence creates a difficulty for adversaries to attack by using the gradient of the classifier. After detecting adversarial examples, we show that many of them can be recovered by simply performing a small average filter on the image. Those findings should lead to more insights about the classification mechanisms in deep convolutional neural networks.

This paper presents a novel approach to ranking unsupervised video segment proposals. This is important in order to obtain a small set of object proposals that cover diverse objects in the video. This paper utilizes a convolutional-recurrent network framework for learning a score function of video segments. A modified Determinantal Point Processes (DPP) approach is used on the results to sample a diverse and meaningful set of objects from the video. Given a pool of video object proposals, our approach can rapidly find a small and diverse set of objects which could then be used for subsequent processing. Results on the challenging VSB-100 dataset demonstrate the performance of our approach.

We propose a robust algorithm to generate video segment proposals. The proposals generated by our method can start from any frame in the video and are robust to complete occlusions. Our method does not assume specific motion models and even has a limited capability to generalize across videos. We build on our previous least squares tracking framework, where image segment proposals are generated and tracked using learned appearance models. The innovation in our new method lies in the use of two efficient moves, the merge move and free addition, to efficiently start segments from any frame and track them through complete occlusions, without much additional computation. Segment size interpolation is used for effectively detecting occlusions. We propose a new metric for evaluating video segment proposals on the challenging VSB-100 benchmark and present state-of-the-art results. Preliminary results are also shown for the potential use of our framework to track segments across different videos.

Object proposals have recently fueled the progress in detection performance. These proposals aim to provide category-agnostic localizations for all objects in an image. One way to generate proposals is to perform parametric min-cuts over seed locations. This paper demonstrates that standard parametric-cut models are ineffective in obtaining medium-sized objects, which we refer to as the middle child problem. We propose a new energy minimization framework incorporating geodesic distances between segments which solves this problem. In addition, we introduce a new superpixel merging algorithm which can generate a small set of seeds that reliably cover a large number of objects of all sizes. We call our method POISE - "Proposals for Objects from Improved Seeds and Energies." POISE enables parametric min-cuts to reach their full potential. On PASCAL VOC it generates ~2,640 segments with an average overlap of 0.81, whereas the closest competing methods require more than 4,200 proposals to reach the same accuracy. We show detailed quantitative comparisons against 5 state-of-the-art methods on PASCAL VOC and Microsoft COCO segmentation challenges.

This paper revisits the classical multiple hypotheses tracking (MHT) algorithm in a tracking-by-detection framework. The success of MHT largely depends on the ability to maintain a small list of potential hypotheses, which can be facilitated with the accurate object detectors that are currently available. We demonstrate that a classical MHT implementation from the 90's can come surprisingly close to the performance of state-of-the-art methods on standard benchmark datasets. In order to further utilize the strength of MHT in exploiting higher-order information, we introduce a method for training online appearance models for each track hypothesis. We show that appearance models can be learned efficiently via a regularized least squares framework, requiring only a few extra operations for each hypothesis branch. We obtain state-of-the-art results on popular tracking-by-detection datasets such as PETS and the recent MOT challenge.

The Continuous-Time Hidden Markov Model (CT-HMM) is an attractive approach to modeling disease progression due to its ability to describe noisy observations arriving irregularly in time. However, the lack of an efficient parameter learning algorithm for CT-HMM restricts its use to very small models or requires unrealistic constraints on the state transitions. In this paper, we present the first complete characterization of efficient EM-based learning methods for CT-HMM models. We demonstrate that the learning problem consists of two challenges: the estimation of posterior state probabilities and the computation of end-state conditioned statistics. We solve the first challenge by reformulating the estimation problem in terms of an equivalent discrete time-inhomogeneous hidden Markov model. The second challenge is addressed by adapting three approaches from the continuous time Markov chain literature to the CT-HMM domain. We demonstrate the use of CT-HMMs with more than 100 states to visualize and predict disease progression using a glaucoma dataset and an Alzheimer's disease dataset.

Sentiment classification is a topic-sensitive task, i.e., a classifier trained from one topic will perform worse on another. This is especially a problem for the tweets sentiment analysis. Since the topics in Twitter are very diverse, it is impossible to train a universal classifier for all topics. Moreover, compared to product review, Twitter lacks data labeling and a rating mechanism to acquire sentiment labels. The extremely sparse text of tweets also brings down the performance of a sentiment classifier. In this paper, we propose a semi-supervised topic-adaptive sentiment classification (TASC) model, which starts with a classifier built on common features and mixed labeled data from various topics. It minimizes the hinge loss to adapt to unlabeled data and features including topic-related sentiment words, authors' sentiments and sentiment connections derived from“@” mentions of tweets, named as topic-adaptive features. Text and non-text features are extracted and naturally split into two views for co-training. The TASC learning algorithm updates topic-adaptive features based on the collaborative selection of unlabeled data, which in turn helps to select more reliable tweets to boost the performance. We also design the adapting model along a timeline (TASC-t) for dynamic tweets. An experiment on 6 topics from published tweet corpuses demonstrates that TASC outperforms other well-known supervised and ensemble classifiers. It also beats those semi-supervised learning methods without feature adaption. Meanwhile, TASC-t can also achieve impressive accuracy and F-score. Finally, with timeline visualization of “river” graph, people can intuitively grasp the ups and downs of sentiments' evolvement, and the intensity by color gradation.

A fundamental challenge to sensory processing tasks in perception and robotics is the problem of obtaining data associations across views. We present a robust solution for ascertaining potentially dense surface patch (super pixel) associations, requiring just range information. Our approach involves decomposition of a view into regularized surface patches. We represent them as sequences expressing geometry invariantly over their super pixel neighborhoods. Uniquely consistent partial orderings are utilized for the same. We match these representations robustly through an optimal sequence comparison metric based on the Damerau-Levenshtein distance. The approach is able to perform under wide baselines, heavy rotations, significant occlusions and partially overlapping content. The technique does not require any priors -- motion or otherwise, and does not make restrictive assumptions on scene structure and sensor movement. It does not require appearance -- is hence more widely applicable than appearance reliant methods, and invulnerable to related ambiguities such as texture less or aliased content. We present promising qualitative and quantitative results under diverse settings, along with comparatives with popular approaches based on range as well as RGB-D data.

Popular figure-ground segmentation algorithms generate a pool of boundary-aligned segment proposals that can be used in subsequent object recognition engines. These algorithms can recover most image objects with high accuracy, but are usually computationally intensive since many graph cuts are computed with different enumerations of segment seeds. In this paper we propose an algorithm, RIGOR, for efficiently generating a pool of overlapping segment proposals in images. By precomputing a graph which can be used for parametric min-cuts over different seeds, we speed up the generation of the segment pool. In addition, we have made design choices that avoid extensive computations without losing performance. In particular, we demonstrate that the segmentation performance of our algorithm is slightly better than the state-of-the-art on the PASCAL VOC dataset, while being an order of magnitude faster.

Sentiment analysis predicts the presence of positive or negative emotions in a text document. In this paper we consider higher dimensional extensions of the sentiment concept, which represent a richer set of human emotions. Our approach goes beyond previous work in that our model contains a continuous manifold rather than a finite set of human emotions. We investigate the resulting model, compare it to psychological observations, and explore its predictive capabilities. Besides obtaining significant improvements over a baseline without manifold, we are also able to visualize different notions of positive sentiment in different domains.

Archives of human rights violations reports, by virtue of their poor metadata, basis in natural language, and scale, obscure fine grain analyses of violation event patterns. Cross-document coreference of victim or perpetrator occurrences from across a corpus is challenging, particularly when those mentions relate to different events. These challenges are emblematic of the transition from small scale to big data analysis in the humanities. This paper discusses these issues and proposes a framework to address these challenges so as to explore narrative construction and the formation of collective memory. Though our framework is based on processing human rights violation reports, it can be readily extended to support other big data problems in the humanities.

Sentiment classification is an important problem in tweets mining. There lack labeled data and rating mechanism for generating them in Twitter service. And topics in Twitter are more diverse while sentiment classifiers always dedicate themselves to a specific domain or topic. Thus it is a challenge to make sentiment classification adaptive to diverse topics without sufficient labeled data. Therefore we formally propose an adaptive multiclass SVM model which transfers an initial common sentiment classifier to a topic-adaptive one. To tackle the tweet sparsity, non-text features are explored besides the conventional text features, which are intuitively split into two views. An iterative algorithm is proposed for solving this model by alternating among three steps: optimization, unlabeled data selection and adaptive feature expansion steps. The algorithm alternatively minimizes the margins of two independent objectives on different views to learn coefficient matrices, which are collaboratively used for unlabeled tweets selection from the topic that the algorithm is adapting to. And then topic-adaptive sentiment words are expended based on the above selection, in turn to help the first two steps find more confident and unlabeled tweets and boost the final performance. Comparing with the well-known supervised sentiment classifiers and semi-supervised approaches, our algorithm achieves promising increases in accuracy averagely on the 6 topics from public tweet corpus.

In this paper we present an inference procedure for the semantic segmentation of images. Different from many CRF approaches that rely on dependencies modeled with unary and pairwise pixel or super pixel potentials, our method is entirely based on estimates of the overlap between each of a set of mid-level object segmentation proposals and the objects present in the image. We define continuous latent variables on super pixels obtained by multiple intersections of segments, then output the optimal segments from the inferred super pixel statistics. The algorithm is capable of recombine and refine initial mid-level proposals, as well as handle multiple interacting objects, even from the same class, all in a consistent joint inference framework by maximizing the composite likelihood of the underlying statistical model using an EM algorithm. In the PASCAL VOC segmentation challenge, the proposed approach obtains high accuracy and successfully handles images of complex object interactions.

We propose an unsupervised video segmentation approach by simultaneously tracking multiple holistic figure-ground segments. Segment tracks are initialized from a pool of segment proposals generated from a figure-ground segmentation algorithm. Then, online non-local appearance models are trained incrementally for each track using a multi-output regularized least squares formulation. By using the same set of training examples for all segment tracks, a computational trick allows us to track hundreds of segment tracks efficiently, as well as perform optimal online updates in closed-form. Besides, a new composite statistical inference approach is proposed for refining the obtained segment tracks, which breaks down the initial segment proposals and recombines for better ones by utilizing high-order statistic estimates from the appearance model and enforcing temporal consistency. For evaluating the algorithm, a dataset, SegTrack v2, is collected with about 1,000 frames with pixel-level annotations. The proposed framework outperforms state-of-the-art approaches in the dataset, showing its efficiency and robustness to challenges in different video sequences.

For cold-start recommendation, it is important to rapidly profile new users and generate a good initial set of recommendations through an interview process --- users should be queried adaptively in a sequential fashion, and multiple items should be offered for opinion solicitation at each trial. In this work, we propose a novel algorithm that learns to conduct the interview process guided by a decision tree with multiple questions at each split. The splits, represented as sparse weight vectors, are learned through an L_1-constrained optimization framework. The users are directed to child nodes according to the inner product of their responses and the corresponding weight vector. More importantly, to account for the variety of responses coming to a node, a linear regressor is learned within each node using all the previously obtained answers as input to predict item ratings. A user study, preliminary but first in its kind in cold-start recommendation, is conducted to explore the efficient number and format of questions being asked in a recommendation survey to minimize user cognitive efforts. Quantitative experimental validations also show that the proposed algorithm outperforms state-of-the-art approaches in terms of both the prediction accuracy and user cognitive efforts.

The random Fourier embedding methodology can be used to approximate the performance of non-linear kernel classifiers in linear time on the number of training examples. However, there still exists a non-trivial performance gap between the approximation and the nonlinear models, especially for the exponential χ2kernel, one of the most powerful models for histograms. Based on analogies with Chebyshev polynomials, we propose an asymptotically convergent analytic series of the χ2measure. The new series removes the need to use periodic approximations to the χ2function, as typical in previous methods, and improves the classification accuracy when used in the random Fourier approximation of the exponential χ2kernel. Besides, out-of-core principal component analysis (PCA) methods are introduced to reduce the dimensionality of the approximation and achieve better performance at the expense of only an additional constant factor to the time complexity. Moreover, when PCA is performed jointly on the training and unlabeled testing data, further performance improvements can be obtained. The proposed approaches are tested on the PASCAL VOC 2010 segmentation and the ImageNet ILSVR-C 2010 datasets, and shown to give statistically significant improvements over alternative approximation methods.

We present an approach for automatic 3D human pose reconstruction from monocular images, based on a discriminative formulation with latent segmentation inputs. We advanced the field of structured prediction and human pose reconstruction on several fronts. First, by working with a pool of figure-ground segment hypotheses, the prediction problem is formulated in terms of combined learning and inference over segment hypotheses and 3D human articular configurations. Beside constructing tractable formulations for the combined segment selection and pose estimation problem, we propose new augmented kernels that can better encode complex dependencies between output variables. Furthermore, we provide primal linear re-formulations based on Fourier kernel approximations, in order to scale-up the non-linear latent structured prediction methodology. The proposed models are shown to be competitive in the HumanEva benchmark and are also illustrated in a clip collected from a Hollywood movie, where the model can infer human poses from monocular images captured in complex environments.

We present an approach to visual object-class recognition and segmentation based on a pipeline that combines multiple, holistic figure-ground hypotheses generated in a bottom-up, object independent process. Decisions are performed based on continuous estimates of the spatial overlap between image segment hypotheses and each putative class. We differ from existing approaches not only in our seemingly unreasonable assumption that good object-level segments can be obtained in a feed-forward fashion, but also in framing recognition as a regression problem. Instead of focusing on a one-vs-all winning margin that can scramble ordering inside the non-maximum (non-winning) set, learning produces a globally consistent ranking with close ties to segment quality, hence to the extent entire object or part hypotheses spatially overlap with the ground truth. We demonstrate results beyond the current state of the art for image classification, object detection and semantic segmentation, in a number of challenging datasets including Caltech-101, ETHZ-Shape and PASCAL VOC 2009.

Multiple-Instance learning has been long known as a hard non-convex problem. In this work, we propose an approach that recasts it as a convex likelihood ratio estimation problem. Firstly, the constraint in multiple-instance learning is reformulated into a convex constraint on the likelihood ratio. Then we show that a joint estimation of a likelihood ratio function and the likelihood on training instances can be learned convexly. Theoretically, we prove a quantitative relationship between the risk estimated under the 0-1 classification loss, and under a loss function for likelihood ratio estimation. It is shown that our likelihood ratio estimation is generally a good surrogate for the 0-1 loss, and separates positive and negative instances well. However with the joint estimation it tends to underestimate the likelihood of an example to be positive. We propose to use these likelihood ratio estimates as features, and learn a linear combination on them to classify the bags. Experiments on synthetic and real datasets show the superiority of the approach.

Distance metric learning and nonlinear dimensionality reduction are two interesting and active topics in recent years. However, the connection between them is not thoroughly studied yet. In this paper, a transductive framework of distance metric learning is proposed and its close connection with many nonlinear spectral dimensionality reduction methods is elaborated. Furthermore, we prove a representer theorem for our framework, linking it with function estimation in an RKHS, and making it possible for generalization to unseen test samples. In our framework, it suffices to solve a sparse eigenvalue problem, thus datasets with 105 samples can be handled. Finally, experiment results on synthetic data, several UCI databases and the MNIST handwritten digit database are shown.

We propose a nonlinear dimensionality reduction algorithm called partitional local tangent space alignment (PLTSA), which is based on VQPCA and LTSA. In the algorithm, the sample space is first divided into overlapped blocks by the X-means algorithm. Then each point is projected to the local tangent space of the block to which the point belongs, to get its local low-dimensional coordinate. The global low-dimensional embedded manifold is obtained from local coordinates via local affine transformations. PLTSA is a better-scaled algorithm than LTSA, in that it provides a means of mapping newcome data with much smaller time and space requirements, and works on a much smaller optimization matrix. Since it gives the global coordinates of the data, it is better than VQPCA. The performance of PLTSA is illustrated by results on surfaces in 3D Euclidean spaces and MNIST database.