Contrastive Learning and Self-Training for Unsupervised Domain Adaptation in Semantic Segmentation

Deep convolutional neural networks have considerably improved state-of-the-art results for semantic segmentation. Nevertheless, even modern architectures lack the ability to generalize well to a test dataset that originates from a different domain. To avoid the costly annotation of training data for unseen domains, unsupervised domain adaptation (UDA) attempts to provide efficient knowledge transfer from a labeled source domain to an unlabeled target domain. Previous work has mainly focused on minimizing the discrepancy between the two domains by using adversarial training or self-training. While adversarial training may fail to align the correct semantic categories as it minimizes the discrepancy between the global distributions, self-training raises the question of how to provide reliable pseudo-labels. To align the correct semantic categories across domains, we propose a contrastive learning approach that adapts category-wise centroids across domains. Furthermore, we extend our method with self-training, where we use a memory-efficient temporal ensemble to generate consistent and reliable pseudo-labels. Although both contrastive learning and self-training (CLST) through temporal ensembling enable knowledge transfer between two domains, it is their combination that leads to a symbiotic structure. We validate our approach on two domain adaptation benchmarks: GTA5 $\rightarrow$ Cityscapes and SYNTHIA $\rightarrow$ Cityscapes. Our method achieves better or comparable results than the state-of-the-art. We will make the code publicly available.

MT3: Meta Test-Time Training for Self-Supervised Test-Time Adaption

An unresolved problem in Deep Learning is the ability of neural networks to cope with domain shifts during test-time, imposed by commonly fixing network parameters after training. Our proposed method Meta Test-Time Training (MT3), however, breaks this paradigm and enables adaption at test-time. We combine meta-learning, self-supervision and test-time training to learn to adapt to unseen test distributions. By minimizing the self-supervised loss, we learn task-specific model parameters for different tasks. A meta-model is optimized such that its adaption to the different task-specific models leads to higher performance on those tasks. During test-time a single unlabeled image is sufficient to adapt the meta-model parameters. This is achieved by minimizing only the self-supervised loss component resulting in a better prediction for that image. Our approach significantly improves the state-of-the-art results on the CIFAR-10-Corrupted image classification benchmark. Our implementation is available on GitHub.

Graph Attention Recurrent Neural Networks for Correlated Time Series Forecasting -- Full version

We consider a setting where multiple entities inter-act with each other over time and the time-varying statuses of the entities are represented as multiple correlated time series. For example, speed sensors are deployed in different locations in a road network, where the speed of a specific location across time is captured by the corresponding sensor as a time series, resulting in multiple speed time series from different locations, which are often correlated. To enable accurate forecasting on correlated time series, we proposes graph attention recurrent neural networks.First, we build a graph among different entities by taking into account spatial proximity and employ a multi-head attention mechanism to derive adaptive weight matrices for the graph to capture the correlations among vertices (e.g., speeds at different locations) at different timestamps. Second, we employ recurrent neural networks to take into account temporal dependency while taking into account the adaptive weight matrices learned from the first step to consider the correlations among time series.Experiments on a large real-world speed time series data set suggest that the proposed method is effective and outperforms the state-of-the-art in most settings. This manuscript provides a full version of a workshop paper [1].

Uncertainty-Based Biological Age Estimation of Brain MRI Scans

Age is an essential factor in modern diagnostic procedures. However, assessment of the true biological age (BA) remains a daunting task due to the lack of reference ground-truth labels. Current BA estimation approaches are either restricted to skeletal images or rely on non-imaging modalities that yield a whole-body BA assessment. However, various organ systems may exhibit different aging characteristics due to lifestyle and genetic factors. In this initial study, we propose a new framework for organ-specific BA estimation utilizing 3D magnetic resonance image (MRI) scans. As a first step, this framework predicts the chronological age (CA) together with the corresponding patient-dependent aleatoric uncertainty. An iterative training algorithm is then utilized to segregate atypical aging patients from the given population based on the predicted uncertainty scores. In this manner, we hypothesize that training a new model on the remaining population should approximate the true BA behavior. We apply the proposed methodology on a brain MRI dataset containing healthy individuals as well as Alzheimer's patients. We demonstrate the correlation between the predicted BAs and the expected cognitive deterioration in Alzheimer's patients.

Anomaly Detection Based on Selection and Weighting in Latent Space

With the high requirements of automation in the era of Industry 4.0, anomaly detection plays an increasingly important role in higher safety and reliability in the production and manufacturing industry. Recently, autoencoders have been widely used as a backend algorithm for anomaly detection. Different techniques have been developed to improve the anomaly detection performance of autoencoders. Nonetheless, little attention has been paid to the latent representations learned by autoencoders. In this paper, we propose a novel selection-and-weighting-based anomaly detection framework called SWAD. In particular, the learned latent representations are individually selected and weighted. Experiments on both benchmark and real-world datasets have shown the effectiveness and superiority of SWAD. On the benchmark datasets, the SWAD framework has reached comparable or even better performance than the state-of-the-art approaches.

Condensed Composite Memory Continual Learning

Deep Neural Networks (DNNs) suffer from a rapid decrease in performance when trained on a sequence of tasks where only data of the most recent task is available. This phenomenon, known as catastrophic forgetting, prevents DNNs from accumulating knowledge over time. Overcoming catastrophic forgetting and enabling continual learning is of great interest since it would enable the application of DNNs in settings where unrestricted access to all the training data at any time is not always possible, e.g. due to storage limitations or legal issues. While many recently proposed methods for continual learning use some training examples for rehearsal, their performance strongly depends on the number of stored examples. In order to improve performance of rehearsal for continual learning, especially for a small number of stored examples, we propose a novel way of learning a small set of synthetic examples which capture the essence of a complete dataset. Instead of directly learning these synthetic examples, we learn a weighted combination of shared components for each example that enables a significant increase in memory efficiency. We demonstrate the performance of our method on commonly used datasets and compare it to recently proposed related methods and baselines.

SLPC: a VRNN-based approach for stochastic lidar prediction and completion in autonomous driving

Predicting future 3D LiDAR pointclouds is a challenging task that is useful in many applications in autonomous driving such as trajectory prediction, pose forecasting and decision making. In this work, we propose a new LiDAR prediction framework that is based on generative models namely Variational Recurrent Neural Networks (VRNNs), titled Stochastic LiDAR Prediction and Completion (SLPC). Our algorithm is able to address the limitations of previous video prediction frameworks when dealing with sparse data by spatially inpainting the depth maps in the upcoming frames. Our contributions can thus be summarized as follows: we introduce the new task of predicting and completing depth maps from spatially sparse data, we present a sparse version of VRNNs and an effective self-supervised training method that does not require any labels. Experimental results illustrate the effectiveness of our framework in comparison to the state of the art methods in video prediction.

End-to-end Interpretable Neural Motion Planner

In this paper, we propose a neural motion planner (NMP) for learning to drive autonomously in complex urban scenarios that include traffic-light handling, yielding, and interactions with multiple road-users. Towards this goal, we design a holistic model that takes as input raw LIDAR data and a HD map and produces interpretable intermediate representations in the form of 3D detections and their future trajectories, as well as a cost volume defining the goodness of each position that the self-driving car can take within the planning horizon. We then sample a set of diverse physically possible trajectories and choose the one with the minimum learned cost. Importantly, our cost volume is able to naturally capture multi-modality. We demonstrate the effectiveness of our approach in real-world driving data captured in several cities in North America. Our experiments show that the learned cost volume can generate safer planning than all the baselines.

PLUME: Efficient 3D Object Detection from Stereo Images

3D object detection plays a significant role in various robotic applications including self-driving. While many approaches rely on expensive 3D sensors like LiDAR to produce accurate 3D estimates, stereo-based methods have recently shown promising results at a lower cost. Existing methods tackle the problem in two steps: first depth estimation is performed, a pseudo LiDAR point cloud representation is computed from the depth estimates, and then object detection is performed in 3D space. However, because the two separate tasks are optimized in different metric spaces, the depth estimation is biased towards big objects and may cause sub-optimal performance of 3D detection. In this paper we propose a model that unifies these two tasks in the same metric space for the first time. Specifically, our model directly constructs a pseudo LiDAR feature volume (PLUME) in 3D space, which is used to solve both occupancy estimation and object detection tasks. PLUME achieves state-of-the-art performance on the challenging KITTI benchmark, with significantly reduced inference time compared with existing methods.

Auto4D: Learning to Label 4D Objects from Sequential Point Clouds

In the past few years we have seen great advances in 3D object detection thanks to deep learning methods. However, they typically rely on large amounts of high-quality labels to achieve good performance, which often require time-consuming and expensive work by human annotators. To address this we propose an automatic annotation pipeline that generates accurate object trajectories in 3D (ie, 4D labels) from LiDAR point clouds. Different from previous works that consider single frames at a time, our approach directly operates on sequential point clouds to combine richer object observations. The key idea is to decompose the 4D label into two parts: the 3D size of the object, and its motion path describing the evolution of the object's pose through time. More specifically, given a noisy but easy-to-get object track as initialization, our model first estimates the object size from temporally aggregated observations, and then refines its motion path by considering both frame-wise observations as well as temporal motion cues. We validate the proposed method on a large-scale driving dataset and show that our approach achieves significant improvements over the baselines. We also showcase the benefits of our approach under the annotator-in-the-loop setting.