In this paper, we analyze different methods to mitigate inherent geographical biases present in state of the art image classification models. We first quantitatively present this bias in two datasets - The Dollar Street Dataset and ImageNet, using images with location information. We then present different methods which can be employed to reduce this bias. Finally, we analyze the effectiveness of the different techniques on making these models more robust to geographical locations of the images.
相關內容
Purpose: To develop a method for automated segmentation of hypothalamus subregions informed by ultra-high resolution ex vivo magnetic resonance images (MRI), which generalizes across MRI sequences and resolutions without retraining. Materials and Methods: We trained our deep learning method, H-synEx, with synthetic images derived from label maps built from ultra-high resolution ex vivo MRI scans, which enables finer-grained manual segmentation when compared with 1mm isometric in vivo images. We validated this retrospective study using 1535 in vivo images from six datasets and six MRI sequences. The quantitative evaluation used the Dice Coefficient (DC) and Average Hausdorff distance (AVD). Statistical analysis compared hypothalamic subregion volumes in controls, Alzheimer's disease (AD), and behavioral variant frontotemporal dementia (bvFTD) subjects using the area under the curve (AUC) and Wilcoxon rank sum test. Results: H-SynEx can segment the hypothalamus across various MRI sequences, encompassing FLAIR sequences with significant slice spacing (5mm). Using hypothalamic volumes on T1w images to distinguish control from AD and bvFTD patients, we observed AUC values of 0.74 and 0.79 respectively. Additionally, AUC=0.66 was found for volume variation on FLAIR scans when comparing control and non-patients. Conclusion: Our results show that H-SynEx successfully leverages information from ultra-high resolution scans to segment in vivo from different MRI sequences such as T1w, T2w, PD, qT1, FA, and FLAIR. We also found that our automated segmentation was able to discriminate controls versus patients on FLAIR images with 5mm spacing. H-SynEx is openly available at //github.com/liviamarodrigues/hsynex.
In this paper, we establish the partial correlation graph for multivariate continuous-time stochastic processes, assuming only that the underlying process is stationary and mean-square continuous with expectation zero and spectral density function. In the partial correlation graph, the vertices are the components of the process and the undirected edges represent partial correlations between the vertices. To define this graph, we therefore first introduce the partial correlation relation for continuous-time processes and provide several equivalent characterisations. In particular, we establish that the partial correlation relation defines a graphoid. The partial correlation graph additionally satisfies the usual Markov properties and the edges can be determined very easily via the inverse of the spectral density function. Throughout the paper, we compare and relate the partial correlation graph to the mixed (local) causality graph of Fasen-Hartmann and Schenk (2023a). Finally, as an example, we explicitly characterise and interpret the edges in the partial correlation graph for the popular multivariate continuous-time AR (MCAR) processes.
In this paper, we present a novel transformer architecture tailored for learning robust power system state representations, which strives to optimize power dispatch for the power flow adjustment across different transmission sections. Specifically, our proposed approach, named Powerformer, develops a dedicated section-adaptive attention mechanism, separating itself from the self-attention used in conventional transformers. This mechanism effectively integrates power system states with transmission section information, which facilitates the development of robust state representations. Furthermore, by considering the graph topology of power system and the electrical attributes of bus nodes, we introduce two customized strategies to further enhance the expressiveness: graph neural network propagation and multi-factor attention mechanism. Extensive evaluations are conducted on three power system scenarios, including the IEEE 118-bus system, a realistic 300-bus system in China, and a large-scale European system with 9241 buses, where Powerformer demonstrates its superior performance over several baseline methods.
In this paper, we present an exploration and assessment of employing a centralized deep Q-network (DQN) controller as a substitute for the prevalent use of PID controllers in the context of 6DOF swimming robots. Our primary focus centers on illustrating this transition with the specific case of underwater object tracking. DQN offers advantages such as data efficiency and off-policy learning, while remaining simpler to implement than other reinforcement learning methods. Given the absence of a dynamic model for our robot, we propose an RL agent to control this multi-input-multi-output (MIMO) system, where a centralized controller may offer more robust control than distinct PIDs. Our approach involves initially using classical controllers for safe exploration, then gradually shifting to DQN to take full control of the robot. We divide the underwater tracking task into vision and control modules. We use established methods for vision-based tracking and introduce a centralized DQN controller. By transmitting bounding box data from the vision module to the control module, we enable adaptation to various objects and effortless vision system replacement. Furthermore, dealing with low-dimensional data facilitates cost-effective online learning for the controller. Our experiments, conducted within a Unity-based simulator, validate the effectiveness of a centralized RL agent over separated PID controllers, showcasing the applicability of our framework for training the underwater RL agent and improved performance compared to traditional control methods. The code for both real and simulation implementations is at //github.com/FARAZLOTFI/underwater-object-tracking.
In this paper, we provide a theoretical analysis of the recently introduced weakly adversarial networks (WAN) method, used to approximate partial differential equations in high dimensions. We address the existence and stability of the solution, as well as approximation bounds. More precisely, we prove the existence of discrete solutions, intended in a suitable weak sense, for which we prove a quasi-best approximation estimate similar to Cea's lemma, a result commonly found in finite element methods. We also propose two new stabilized WAN-based formulas that avoid the need for direct normalization. Furthermore, we analyze the method's effectiveness for the Dirichlet boundary problem that employs the implicit representation of the geometry. The key requirement for achieving the best approximation outcome is to ensure that the space for the test network satisfies a specific condition, known as the inf-sup condition, essentially requiring that the test network set is sufficiently large when compared to the trial space. The method's accuracy, however, is only determined by the space of the trial network. We also devise a pseudo-time XNODE neural network class for static PDE problems, yielding significantly faster convergence results than the classical DNN network.
In this paper, we develop a class of high-order conservative methods for simulating non-equilibrium radiation diffusion problems. Numerically, this system poses significant challenges due to strong nonlinearity within the stiff source terms and the degeneracy of nonlinear diffusion terms. Explicit methods require impractically small time steps, while implicit methods, which offer stability, come with the challenge to guarantee the convergence of nonlinear iterative solvers. To overcome these challenges, we propose a predictor-corrector approach and design proper implicit-explicit time discretizations. In the predictor step, the system is reformulated into a nonconservative form and linear diffusion terms are introduced as a penalization to mitigate strong nonlinearities. We then employ a Picard iteration to secure convergence in handling the nonlinear aspects. The corrector step guarantees the conservation of total energy, which is vital for accurately simulating the speeds of propagating sharp fronts in this system. For spatial approximations, we utilize local discontinuous Galerkin finite element methods, coupled with positive-preserving and TVB limiters. We validate the orders of accuracy, conservation properties, and suitability of using large time steps for our proposed methods, through numerical experiments conducted on one- and two-dimensional spatial problems. In both homogeneous and heterogeneous non-equilibrium radiation diffusion problems, we attain a time stability condition comparable to that of a fully implicit time discretization. Such an approach is also applicable to many other reaction-diffusion systems.
We propose a novel algorithm for the support estimation of partially known Gaussian graphical models that incorporates prior information about the underlying graph. In contrast to classical approaches that provide a point estimate based on a maximum likelihood or a maximum a posteriori criterion using (simple) priors on the precision matrix, we consider a prior on the graph and rely on annealed Langevin diffusion to generate samples from the posterior distribution. Since the Langevin sampler requires access to the score function of the underlying graph prior, we use graph neural networks to effectively estimate the score from a graph dataset (either available beforehand or generated from a known distribution). Numerical experiments demonstrate the benefits of our approach.
In this paper, we introduce a new shape functional defined for toroidal domains that we call harmonic helicity, and study its shape optimization. Given a toroidal domain, we consider its associated harmonic field. The latter is the magnetic field obtained uniquely up to normalization when imposing zero normal trace and zero electrical current inside the domain. We then study the helicity of this field, which is a quantity of interest in magneto-hydrodynamics corresponding to the L2 product of the field with its image by the Biot--Savart operator. To do so, we begin by discussing the appropriate functional framework and an equivalent PDE characterization. We then focus on shape optimization, and we identify the shape gradient of the harmonic helicity. Finally, we study and implement an efficient numerical scheme to compute harmonic helicity and its shape gradient using finite elements exterior calculus.
In this work, a Generalized Finite Difference (GFD) scheme is presented for effectively computing the numerical solution of a parabolic-elliptic system modelling a bacterial strain with density-suppressed motility. The GFD method is a meshless method known for its simplicity for solving non-linear boundary value problems over irregular geometries. The paper first introduces the basic elements of the GFD method, and then an explicit-implicit scheme is derived. The convergence of the method is proven under a bound for the time step, and an algorithm is provided for its computational implementation. Finally, some examples are considered comparing the results obtained with a regular mesh and an irregular cloud of points.
Typical pipelines for model geometry generation in computational biomedicine stem from images, which are usually considered to be at rest, despite the object being in mechanical equilibrium under several forces. We refer to the stress-free geometry computation as the reference configuration problem, and in this work we extend such a formulation to the theory of fully nonlinear poroelastic media. The main steps are (i) writing the equations in terms of the reference porosity and (ii) defining a time dependent problem whose steady state solution is the reference porosity. This problem can be computationally challenging as it can require several hundreds of iterations to converge, so we propose the use of Anderson acceleration to speed up this procedure. Our evidence shows that this strategy can reduce the number of iterations up to 80\%. In addition, we note that a primal formulation of the nonlinear mass conservation equations is not consistent due to the presence of second order derivatives of the displacement, which we alleviate through adequate mixed formulations. All claims are validated through numerical simulations in both idealized and realistic scenarios.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191