An Analysis of Switchback Designs in Reinforcement Learning

This paper offers a detailed investigation of switchback designs in A/B testing, which alternate between baseline and new policies over time. Our aim is to thoroughly evaluate the effects of these designs on the accuracy of their resulting average treatment effect (ATE) estimators. We propose a novel "weak signal analysis" framework, which substantially simplifies the calculations of the mean squared errors (MSEs) of these ATEs in Markov decision process environments. Our findings suggest that (i) when the majority of reward errors are positively correlated, the switchback design is more efficient than the alternating-day design which switches policies in a daily basis. Additionally, increasing the frequency of policy switches tends to reduce the MSE of the ATE estimator. (ii) When the errors are uncorrelated, however, all these designs become asymptotically equivalent. (iii) In cases where the majority of errors are negative correlated, the alternating-day design becomes the optimal choice. These insights are crucial, offering guidelines for practitioners on designing experiments in A/B testing. Our analysis accommodates a variety of policy value estimators, including model-based estimators, least squares temporal difference learning estimators, and double reinforcement learning estimators, thereby offering a comprehensive understanding of optimal design strategies for policy evaluation in reinforcement learning.

DomainVerse: A Benchmark Towards Real-World Distribution Shifts For Tuning-Free Adaptive Domain Generalization

Traditional cross-domain tasks, including domain adaptation and domain generalization, rely heavily on training model by source domain data. With the recent advance of vision-language models (VLMs), viewed as natural source models, the cross-domain task changes to directly adapt the pre-trained source model to arbitrary target domains equipped with prior domain knowledge, and we name this task Adaptive Domain Generalization (ADG). However, current cross-domain datasets have many limitations, such as unrealistic domains, unclear domain definitions, and the inability to fine-grained domain decomposition, which drives us to establish a novel dataset DomainVerse for ADG. Benefiting from the introduced hierarchical definition of domain shifts, DomainVerse consists of about 0.5 million images from 390 fine-grained realistic domains. With the help of the constructed DomainVerse and VLMs, we propose two methods called Domain CLIP and Domain++ CLIP for tuning-free adaptive domain generalization. Extensive and comprehensive experiments demonstrate the significance of the dataset and the effectiveness of the proposed methods.

Counterfactual Graph Transformer for Traffic Flow Prediction

Traffic flow prediction (TFP) is a fundamental problem of the Intelligent Transportation System (ITS), as it models the latent spatial-temporal dependency of traffic flow for potential congestion prediction. Recent graph-based models with multiple kinds of attention mechanisms have achieved promising performance. However, existing methods for traffic flow prediction tend to inherit the bias pattern from the dataset and lack interpretability. To this end, we propose a Counterfactual Graph Transformer (CGT) model with an instance-level explainer (e.g., finding the important subgraphs) specifically designed for TFP. We design a perturbation mask generator over input sensor features at the time dimension and the graph structure on the graph transformer module to obtain spatial and temporal counterfactual explanations. By searching the optimal perturbation masks on the input data feature and graph structures, we can obtain the concise and dominant data or graph edge links for the subsequent TFP task. After re-training the utilized graph transformer model after counterfactual perturbation, we can obtain improved and interpretable traffic flow prediction. Extensive results on three real-world public datasets show that CGT can produce reliable explanations and is promising for traffic flow prediction.

An Entire Renal Anatomy Extraction Network for Advanced CAD During Partial Nephrectomy

Partial nephrectomy (PN) is common surgery in urology. Digitization of renal anatomies brings much help to many computer-aided diagnosis (CAD) techniques during PN. However, the manual delineation of kidney vascular system and tumor on each slice is time consuming, error-prone, and inconsistent. Therefore, we proposed an entire renal anatomies extraction method from Computed Tomographic Angiographic (CTA) images fully based on deep learning. We adopted a coarse-to-fine workflow to extract target tissues: first, we roughly located the kidney region, and then cropped the kidney region for more detail extraction. The network we used in our workflow is based on 3D U-Net. To dealing with the imbalance of class contributions to loss, we combined the dice loss with focal loss, and added an extra weight to prevent excessive attention. We also improved the manual annotations of vessels by merging semi-trained model's prediction and original annotations under supervision. We performed several experiments to find the best-fitting combination of variables for training. We trained and evaluated the models on our 60 cases dataset with 3 different sources. The average dice score coefficient (DSC) of kidney, tumor, cyst, artery, and vein, were 90.9%, 90.0%, 89.2%, 80.1% and 82.2% respectively. Our modulate weight and hybrid strategy of loss function increased the average DSC of all tissues about 8-20%. Our optimization of vessel annotation improved the average DSC about 1-5%. We proved the efficiency of our network on renal anatomies segmentation. The high accuracy and fully automation make it possible to quickly digitize the personal renal anatomies, which greatly increases the feasibility and practicability of CAD application on urology surgery.

Retinex Image Enhancement Based on Sequential Decomposition With a Plug-and-Play Framework

The Retinex model is one of the most representative and effective methods for low-light image enhancement. However, the Retinex model does not explicitly tackle the noise problem, and shows unsatisfactory enhancing results. In recent years, due to the excellent performance, deep learning models have been widely used in low-light image enhancement. However, these methods have two limitations: i) The desirable performance can only be achieved by deep learning when a large number of labeled data are available. However, it is not easy to curate massive low/normal-light paired data; ii) Deep learning is notoriously a black-box model [1]. It is difficult to explain their inner-working mechanism and understand their behaviors. In this paper, using a sequential Retinex decomposition strategy, we design a plug-and-play framework based on the Retinex theory for simultaneously image enhancement and noise removal. Meanwhile, we develop a convolutional neural network-based (CNN-based) denoiser into our proposed plug-and-play framework to generate a reflectance component. The final enhanced image is produced by integrating the illumination and reflectance with gamma correction. The proposed plug-and-play framework can facilitate both post hoc and ad hoc interpretability. Extensive experiments on different datasets demonstrate that our framework outcompetes the state-of-the-art methods in both image enhancement and denoising.

Secondary complementary balancing compressive imaging with a free-space balanced amplified photodetector

Single-pixel imaging (SPI) has attracted widespread attention because it generally uses a non-pixelated photodetector and a digital micromirror device (DMD) to acquire the object image. Since the modulated patterns seen from two reflection directions of the DMD are naturally complementary, one can apply complementary balanced measurements to greatly improve the measurement signal-to-noise ratio and reconstruction quality. However, the balance between two reflection arms significantly determines the quality of differential measurements. In this work, we propose and demonstrate a simple secondary complementary balancing mechanism to minimize the impact of the imbalance on the imaging system. In our SPI setup, we used a silicon free-space balanced amplified photodetector with 5 mm active diameter which could directly output the difference between two optical input signals in two reflection arms. Both simulation and experimental results have demonstrated that the use of secondary complementary balancing can result in a better cancellation of direct current components of measurements and a better image restoration quality.

Single-pixel imaging based on weight sort of the Hadamard basis

Single-pixel imaging (SPI) is very popular in subsampling applications, but the random measurement matrices it typically uses will lead to measurement blindness as well as difficulties in calculation and storage, and will also limit the further reduction in sampling rate. The deterministic Hadamard basis has become an alternative choice due to its orthogonality and structural characteristics. There is evidence that sorting the Hadamard basis is beneficial to further reduce the sampling rate, thus many orderings have emerged, but their relations remain unclear and lack a unified theory. Given this, here we specially propose a concept named selection history, which can record the Hadamard spatial folding process, and build a model based on it to reveal the formation mechanisms of different orderings and to deduce the mutual conversion relationship among them. Then, a weight ordering of the Hadamard basis is proposed. Both numerical simulation and experimental results have demonstrated that with this weight sort technique, the sampling rate, reconstruction time and matrix memory consumption are greatly reduced in comparison to traditional sorting methods. Therefore, we believe that this method may pave the way for real-time single-pixel imaging.

Policy Evaluation for Temporal and/or Spatial Dependent Experiments in Ride-sourcing Platforms

Policy evaluation based on A/B testing has attracted considerable interest in digital marketing, but such evaluation in ride-sourcing platforms (e.g., Uber and Didi) is not well studied primarily due to the complex structure of their temporal and/or spatial dependent experiments. Motivated by policy evaluation in ride-sourcing platforms, the aim of this paper is to establish causal relationship between platform's policies and outcomes of interest under a switchback design. We propose a novel potential outcome framework based on a temporal varying coefficient decision process (VCDP) model to capture the dynamic treatment effects in temporal dependent experiments. We further characterize the average treatment effect by decomposing it as the sum of direct effect (DE) and indirect effect (IE). We develop estimation and inference procedures for both DE and IE. Furthermore, we propose a spatio-temporal VCDP to deal with spatiotemporal dependent experiments. For both VCDP models, we establish the statistical properties (e.g., weak convergence and asymptotic power) of our estimation and inference procedures. We conduct extensive simulations to investigate the finite-sample performance of the proposed estimation and inference procedures. We examine how our VCDP models can help improve policy evaluation for various dispatching and dispositioning policies in Didi.

Simultaneous reconstruction and displacement estimation for spectral-domain optical coherence elastography

Optical coherence elastography allows the characterization of the mechanical properties of tissues, and can be performed through estimating local displacement maps from subsequent acquisitions of a sample under different loads. This displacement estimation is limited by noise in the images, which can be high in dynamic systems due to the inability to perform long exposures or B-scan averaging. In this work, we propose a framework for simultaneously enhancing both the image quality and displacement map for elastography, by motion compensated denoising with the block-matching and 4D filtering (BM4D) method, followed by a re-estimation of displacement. We adopt the interferometric synthetic aperture microscopy (ISAM) method to enhance the lateral resolution away from the focal plane, and use sub-pixel cross correlation block matching for non-uniform deformation estimation. We validate this approach on data from a commercial spectral domain optical coherence tomography system, whereby we observe an enhancement of both image and displacement accuracy of up to 33% over a standard approach.

Model-based iterative reconstruction for spectral-domain optical coherence tomography

Spectral domain optical coherence tomography (OCT) offers high resolution multidimensional imaging, but generally suffers from defocussing, intensity falloff and shot noise, causing artifacts and image degradation along the imaging depth. In this work, we develop an iterative statistical reconstruction technique, based upon the interferometric synthetic aperture microscopy (ISAM) model with additive noise, to actively compensate for these effects. For the ISAM re-sampling, we use a non uniform FFT with Kaiser-Bessel interpolation, offering efficiency and high accuracy. We then employ an accelerated gradient descent based algorithm, to minimize the negative log-likelihood of the model, and include spatial or wavelet sparsity based penalty functions, to provide appropriate regularization for given image structures. We evaluate our approach with titanium oxide micro-bead and cucumber samples with a commercial spectral domain OCT system, under various subsampling regimes, and demonstrate superior image quality over traditional reconstruction and ISAM methods.