Deep learning models have gained remarkable performance on a variety of image classification tasks. However, many models suffer from limited performance in clinical or medical settings when data are imbalanced. To address this challenge, we propose a medical-knowledge-guided one-class classification approach that leverages domain-specific knowledge of classification tasks to boost the model's performance. The rationale behind our approach is that some existing prior medical knowledge can be incorporated into data-driven deep learning to facilitate model learning. We design a deep learning-based one-class classification pipeline for imbalanced image classification, and demonstrate in three use cases how we take advantage of medical knowledge of each specific classification task by generating additional middle classes to achieve higher classification performances. We evaluate our approach on three different clinical image classification tasks (a total of 8459 images) and show superior model performance when compared to six state-of-the-art methods. All codes of this work will be publicly available upon acceptance of the paper.
Medical image data are usually imbalanced across different classes. One-class classification has attracted increasing attention to address the data imbalance problem by distinguishing the samples of the minority class from the majority class. Previous methods generally aim to either learn a new feature space to map training samples together or to fit training samples by autoencoder-like models. These methods mainly focus on capturing either compact or descriptive features, where the information of the samples of a given one class is not sufficiently utilized. In this paper, we propose a novel deep learning-based method to learn compact features by adding constraints on the bottleneck features, and to preserve descriptive features by training an autoencoder at the same time. Through jointly optimizing the constraining loss and the autoencoder's reconstruction loss, our method can learn more relevant features associated with the given class, making the majority and minority samples more distinguishable. Experimental results on three clinical datasets (including the MRI breast images, FFDM breast images and chest X-ray images) obtains state-of-art performance compared to previous methods.
Time series data generation has drawn increasing attention in recent years. Several generative adversarial network (GAN) based methods have been proposed to tackle the problem usually with the assumption that the targeted time series data are well-formatted and complete. However, real-world time series (RTS) data are far away from this utopia, e.g., long sequences with variable lengths and informative missing data raise intractable challenges for designing powerful generation algorithms. In this paper, we propose a novel generative framework for RTS data - RTSGAN to tackle the aforementioned challenges. RTSGAN first learns an encoder-decoder module which provides a mapping between a time series instance and a fixed-dimension latent vector and then learns a generation module to generate vectors in the same latent space. By combining the generator and the decoder, RTSGAN is able to generate RTS which respect the original feature distributions and the temporal dynamics. To generate time series with missing values, we further equip RTSGAN with an observation embedding layer and a decide-and-generate decoder to better utilize the informative missing patterns. Experiments on the four RTS datasets show that the proposed framework outperforms the previous generation methods in terms of synthetic data utility for downstream classification and prediction tasks.
Location recommendation is defined as to recommend locations (POIs) to users in location-based services. The existing data-driving approaches of location recommendation suffer from the limitation of the implicit modeling of the geographical factor, which may lead to sub-optimal recommendation results. In this work, we address this problem by introducing knowledge-driven solutions. Specifically, we first construct the Urban Knowledge Graph (UrbanKG) with geographical information and functional information of POIs. On the other side, there exist a fact that the geographical factor not only characterizes POIs but also affects user-POI interactions. To address it, we propose a novel method named UKGC. We first conduct information propagation on two sub-graphs to learn the representations of POIs and users. We then fuse two parts of representations by counterfactual learning for the final prediction. Extensive experiments on two real-world datasets verify that our method can outperform the state-of-the-art methods.
Behavioral cloning has proven to be effective for learning sequential decision-making policies from expert demonstrations. However, behavioral cloning often suffers from the causal confusion problem where a policy relies on the noticeable effect of expert actions due to the strong correlation but not the cause we desire. This paper presents Object-aware REgularizatiOn (OREO), a simple technique that regularizes an imitation policy in an object-aware manner. Our main idea is to encourage a policy to uniformly attend to all semantic objects, in order to prevent the policy from exploiting nuisance variables strongly correlated with expert actions. To this end, we introduce a two-stage approach: (a) we extract semantic objects from images by utilizing discrete codes from a vector-quantized variational autoencoder, and (b) we randomly drop the units that share the same discrete code together, i.e., masking out semantic objects. Our experiments demonstrate that OREO significantly improves the performance of behavioral cloning, outperforming various other regularization and causality-based methods on a variety of Atari environments and a self-driving CARLA environment. We also show that our method even outperforms inverse reinforcement learning methods trained with a considerable amount of environment interaction.
Many works have investigated the adversarial attacks or defenses under the settings where a bounded and imperceptible perturbation can be added to the input. However in the real-world, the attacker does not need to comply with this restriction. In fact, more threats to the deep model come from unrestricted adversarial examples, that is, the attacker makes large and visible modifications on the image, which causes the model classifying mistakenly, but does not affect the normal observation in human perspective. Unrestricted adversarial attack is a popular and practical direction but has not been studied thoroughly. We organize this competition with the purpose of exploring more effective unrestricted adversarial attack algorithm, so as to accelerate the academical research on the model robustness under stronger unbounded attacks. The competition is held on the TianChi platform (\url{https://tianchi.aliyun.com/competition/entrance/531853/introduction}) as one of the series of AI Security Challengers Program.
This paper investigates multiuser multi-input single-output downlink symbiotic radio communication systems assisted by an intelligent reflecting surface (IRS). Different from existing methods ideally assuming the secondary user (SU) can jointly decode information symbols from both the access point (AP) and the IRS via multiuser detection, we consider a more practical SU that only non-coherent detection is available. To characterize the non-coherent decoding performance, a practical upper bound of the average symbol error rate (SER) is derived. Subsequently, we jointly optimize the beamformer at the AP and the phase shifts at the IRS to maximize the average sum-rate of the primary system taking into account the maximum tolerable SER constraint for the SU. To circumvent the couplings of variables, we exploit the Schur complement that facilitates the design of a suboptimal beamforming algorithm based on successive convex approximation. Our simulation results show that compared with various benchmark algorithms, the proposed scheme significantly improves the average sum-rate of the primary system, while guaranteeing the decoding performance of the secondary system.
Miyun Reservoir has produced huge benefits in flood control, agricultural irrigation, power generation, aquaculture, tourism, and urban water supply. Accurately water mapping is of great significance to the ecological environment monitoring of the Miyun Reservoir and the management of the South-to-North Water Diversion Project. On the 60th anniversary of the completion of the Miyun Reservoir, we took the Miyun Reservoir as the study area and collected all the Landsat-5 and Landsat-8 remote sensing images from 1984 to 2020 for water mapping. Based on the spectral, topographical and temporal-spatial characteristics of water, we proposed an automated method for long-term researvoir mapping, which can solve the problems caused by cloud, shadow, ice and snow pixels. Moreover, it can also deal with 'the same objects with different spectra' and spectral mixed problems. The overall accuracy is as high as 98.2% for the case with no cloud or snow/ice cover. The landscape division index is introduced to analyze the morphological changes of Miyun Reservoir. Based on the mapping results, we analyzed the changes of Miyun Reservoir from 1984 to 2020 and the driving factors of them.
Establishing retrieval-based dialogue systems that can select appropriate responses from the pre-built index has gained increasing attention from researchers. For this task, the adoption of pre-trained language models (such as BERT) has led to remarkable progress in a number of benchmarks. There exist two common approaches, including cross-encoders which perform full attention over the inputs, and bi-encoders that encode the context and response separately. The former gives considerable improvements in accuracy but is often inapplicable in practice for large-scale retrieval given the cost of the full attention required for each sample at test time. The latter is efficient for billions of indexes but suffers from sub-optimal performance. In this work, we propose to combine the best of both worlds to build a retrieval system. Specifically, we employ a fast bi-encoder to replace the traditional feature-based pre-retrieval model (such as BM25) and set the response re-ranking model as a more complicated architecture (such as cross-encoder). To further improve the effectiveness of our framework, we train the pre-retrieval model and the re-ranking model at the same time via mutual learning, which enables two models to learn from each other throughout the training process. We conduct experiments on two benchmarks and evaluation results demonstrate the efficiency and effectiveness of our proposed framework.