Superpixel-guided Iterative Learning from Noisy Labels for Medical Image Segmentation

Learning segmentation from noisy labels is an important task for medical image analysis due to the difficulty in acquiring highquality annotations. Most existing methods neglect the pixel correlation and structural prior in segmentation, often producing noisy predictions around object boundaries. To address this, we adopt a superpixel representation and develop a robust iterative learning strategy that combines noise-aware training of segmentation network and noisy label refinement, both guided by the superpixels. This design enables us to exploit the structural constraints in segmentation labels and effectively mitigate the impact of label noise in learning. Experiments on two benchmarks show that our method outperforms recent state-of-the-art approaches, and achieves superior robustness in a wide range of label noises. Code is available at https://github.com/gaozhitong/SP_guided_Noisy_Label_Seg.

Comments on Two Cultures: What have changed over 20 years?

Twenty years ago Breiman (2001) called to our attention a significant cultural division in modeling and data analysis between the stochastic data models and the algorithmic models. Out of his deep concern that the statistical community was so deeply and almost exclusively committed to the former, Breiman warned that we were losing our abilities to solve many real-world problems. Breiman was not the first, and certainly not the only statistician, to sound the alarm; we may refer to none other than John Tukey who wrote almost 60 years ago data analysis is intrinsically an empirical science. However, the bluntness and timeliness of Breimans article made it uniquely influential. It prepared us for the data science era and encouraged a new generation of statisticians to embrace a more broadly defined discipline. Some might argue that The cultural division between these two statistical learning frameworks has been growing at a steady pace in recent years, to quote Mukhopadhyay and Wang (2020). In this commentary, we focus on some of the positive changes over the past 20 years and offer an optimistic outlook for our profession.

Learning Implicit Temporal Alignment for Few-shot Video Classification

Few-shot video classification aims to learn new video categories with only a few labeled examples, alleviating the burden of costly annotation in real-world applications. However, it is particularly challenging to learn a class-invariant spatial-temporal representation in such a setting. To address this, we propose a novel matching-based few-shot learning strategy for video sequences in this work. Our main idea is to introduce an implicit temporal alignment for a video pair, capable of estimating the similarity between them in an accurate and robust manner. Moreover, we design an effective context encoding module to incorporate spatial and feature channel context, resulting in better modeling of intra-class variations. To train our model, we develop a multi-task loss for learning video matching, leading to video features with better generalization. Extensive experimental results on two challenging benchmarks, show that our method outperforms the prior arts with a sizable margin on SomethingSomething-V2 and competitive results on Kinetics.

DER: Dynamically Expandable Representation for Class Incremental Learning

We address the problem of class incremental learning, which is a core step towards achieving adaptive vision intelligence. In particular, we consider the task setting of incremental learning with limited memory and aim to achieve better stability-plasticity trade-off. To this end, we propose a novel two-stage learning approach that utilizes a dynamically expandable representation for more effective incremental concept modeling. Specifically, at each incremental step, we freeze the previously learned representation and augment it with additional feature dimensions from a new learnable feature extractor. This enables us to integrate new visual concepts with retaining learned knowledge. We dynamically expand the representation according to the complexity of novel concepts by introducing a channel-level mask-based pruning strategy. Moreover, we introduce an auxiliary loss to encourage the model to learn diverse and discriminate features for novel concepts. We conduct extensive experiments on the three class incremental learning benchmarks and our method consistently outperforms other methods with a large margin.

Learning Context-aware Task Reasoning for Efficient Meta-reinforcement Learning

Despite recent success of deep network-based Reinforcement Learning (RL), it remains elusive to achieve human-level efficiency in learning novel tasks. While previous efforts attempt to address this challenge using meta-learning strategies, they typically suffer from sampling inefficiency with on-policy RL algorithms or meta-overfitting with off-policy learning. In this work, we propose a novel meta-RL strategy to address those limitations. In particular, we decompose the meta-RL problem into three sub-tasks, task-exploration, task-inference and task-fulfillment, instantiated with two deep network agents and a task encoder. During meta-training, our method learns a task-conditioned actor network for task-fulfillment, an explorer network with a self-supervised reward shaping that encourages task-informative experiences in task-exploration, and a context-aware graph-based task encoder for task inference. We validate our approach with extensive experiments on several public benchmarks and the results show that our algorithm effectively performs exploration for task inference, improves sample efficiency during both training and testing, and mitigates the meta-overfitting problem.

Geometry-aware Deep Network for Single-Image Novel View Synthesis

This paper tackles the problem of novel view synthesis from a single image. In particular, we target real-world scenes with rich geometric structure, a challenging task due to the large appearance variations of such scenes and the lack of simple 3D models to represent them. Modern, learning-based approaches mostly focus on appearance to synthesize novel views and thus tend to generate predictions that are inconsistent with the underlying scene structure. By contrast, in this paper, we propose to exploit the 3D geometry of the scene to synthesize a novel view. Specifically, we approximate a real-world scene by a fixed number of planes, and learn to predict a set of homographies and their corresponding region masks to transform the input image into a novel view. To this end, we develop a new region-aware geometric transform network that performs these multiple tasks in a common framework. Our results on the outdoor KITTI and the indoor ScanNet datasets demonstrate the effectiveness of our network in generating high quality synthetic views that respect the scene geometry, thus outperforming the state-of-the-art methods.

Learning Dynamic Hierarchical Models for Anytime Scene Labeling

With increasing demand for efficient image and video analysis, test-time cost of scene parsing becomes critical for many large-scale or time-sensitive vision applications. We propose a dynamic hierarchical model for anytime scene labeling that allows us to achieve flexible trade-offs between efficiency and accuracy in pixel-level prediction. In particular, our approach incorporates the cost of feature computation and model inference, and optimizes the model performance for any given test-time budget by learning a sequence of image-adaptive hierarchical models. We formulate this anytime representation learning as a Markov Decision Process with a discrete-continuous state-action space. A high-quality policy of feature and model selection is learned based on an approximate policy iteration method with action proposal mechanism. We demonstrate the advantages of our dynamic non-myopic anytime scene parsing on three semantic segmentation datasets, which achieves $90%$ of the state-of-the-art performances by using $15%$ of their overall costs.

Semantic-Aware Depth Super-Resolution in Outdoor Scenes

While depth sensors are becoming increasingly popular, their spatial resolution often remains limited. Depth super-resolution therefore emerged as a solution to this problem. Despite much progress, state-of-the-art techniques suffer from two drawbacks: (i) they rely on the assumption that intensity edges coincide with depth discontinuities, which, unfortunately, is only true in controlled environments; and (ii) they typically exploit the availability of high-resolution training depth maps, which can often not be acquired in practice due to the sensors limitations. By contrast, here, we introduce an approach to performing depth super-resolution in more challenging conditions, such as in outdoor scenes. To this end, we first propose to exploit semantic information to better constrain the super-resolution process. In particular, we design a co-sparse analysis model that learns filters from joint intensity, depth and semantic information. Furthermore, we show how low-resolution training depth maps can be employed in our learning strategy. We demonstrate the benefits of our approach over state-of-the-art depth super-resolution methods on two outdoor scene datasets.

Structured Depth Prediction in Challenging Monocular Video Sequences

In this paper, we tackle the problem of estimating the depth of a scene from a monocular video sequence. In particular, we handle challenging scenarios, such as non-translational camera motion and dynamic scenes, where traditional structure from motion and motion stereo methods do not apply. To this end, we first study the problem of depth estimation from a single image. In this context, we exploit the availability of a pool of images for which the depth is known, and formulate monocular depth estimation as a discrete-continuous optimization problem, where the continuous variables encode the depth of the superpixels in the input image, and the discrete ones represent relationships between neighboring superpixels. The solution to this discrete-continuous optimization problem is obtained by performing inference in a graphical model using particle belief propagation. To handle video sequences, we then extend our single image model to a two-frame one that naturally encodes short-range temporal consistency and inherently handles dynamic objects. Based on the prediction of this model, we then introduce a fully-connected pairwise CRF that accounts for longer range spatio-temporal interactions throughout a video. We demonstrate the effectiveness of our model in both the indoor and outdoor scenarios.

Globally adaptive quantile regression with ultra-high dimensional data

Quantile regression has become a valuable tool to analyze heterogeneous covaraite-response associations that are often encountered in practice. The development of quantile regression methodology for high-dimensional covariates primarily focuses on examination of model sparsity at a single or multiple quantile levels, which are typically pre-specified ad hoc by the users. The resulting models may be sensitive to the specific choices of the quantile levels, leading to difficulties in interpretation and erosion of confidence in the results. In this article, we propose a new penalization framework for quantile regression in the high-dimensional setting. We employ adaptive L1 penalties, and more importantly, propose a uniform selector of the tuning parameter for a set of quantile levels to avoid some of the potential problems with model selection at individual quantile levels. Our proposed approach achieves consistent shrinkage of regression quantile estimates across a continuous range of quantiles levels, enhancing the flexibility and robustness of the existing penalized quantile regression methods. Our theoretical results include the oracle rate of uniform convergence and weak convergence of the parameter estimators. We also use numerical studies to confirm our theoretical findings and illustrate the practical utility of our proposal