Forward Pass Research Articles

Aleatoric and epistemic uncertainty extraction of patient-specific deep learning-based dose predictions in LDR prostate brachytherapy.

Objective.In brachytherapy, deep learning (DL) algorithms have shown the capability of predicting 3D dose volumes. The reliability and accuracy of such methodologies remain under scrutiny for prospective clinical applications. This study aims to establish fast DL-based predictive dose algorithms for low-dose rate (LDR) prostate brachytherapy and to evaluate their uncertainty and stability.Approach.Data from 200 prostate patients, treated with125I sources, was collected. The Monte Carlo (MC) ground truth dose volumes were calculated with TOPAS considering the interseed effects and an organ-based material assignment. Two 3D convolutional neural networks, UNet and ResUNet TSE, were trained using the patient geometry and the seed positions as the input data. The dataset was randomly split into training (150), validation (25) and test (25) sets. The aleatoric (associated with the input data) and epistemic (associated with the model) uncertainties of the DL models were assessed.Main results.For the full test set, with respect to the MC reference, the predicted prostateD90metric had mean differences of -0.64% and 0.08% for the UNet and ResUNet TSE models, respectively. In voxel-by-voxel comparisons, the average global dose difference ratio in the [-1%, 1%] range included 91.0% and 93.0% of voxels for the UNet and the ResUNet TSE, respectively. One forward pass or prediction took 4 ms for a 3D dose volume of 2.56 M voxels (128 × 160 × 128). The ResUNet TSE model closely encoded the well-known physics of the problem as seen in a set of uncertainty maps. The ResUNet TSE rectum D2cchad the largest uncertainty metric of 0.0042.Significance.The proposed DL models serve as rapid dose predictors that consider the patient anatomy and interseed attenuation effects. The derived uncertainty is interpretable, highlighting areas where DL models may struggle to provide accurate estimations. The uncertainty analysis offers a comprehensive evaluation tool for dose predictor model assessment.

CFGPFSR: A Generative Method Combining Facial and GAN Priors for Face Super-Resolution

In recent years, facial prior has been widely applied to enhance the quality of super-resolution (SR) facial images in face super-resolution (FSR) methods based on deep learning. However, most of the existing facial prior-based FSR methods have insufficient attention to local texture details, which can cause the generated SR facial images with overly smooth and unrealistic texture details, and show obvious artifacts under large magnification. With the help of GAN prior, recent advances can produce excellent results in terms of fidelity and realness. A generative framework for FSR is proposed in this work, which combines GAN and facial prior, termed CFGPFSR. Firstly, we pre-train a face StyleGAN2 and a face parsing network (FPN) that can generate decent parsing maps, in which the proposed CFGPFSR exploits rich and varied priors encapsulated in the face StyleGAN2 (GAN prior) and face parsing maps extracted from the FPN (facial prior) for FSR. Moreover, we introduce the Channel-Split Spatial Feature Transform (CS-SFT) method to further improve FSR performance. GAN and facial priors are introduced into the FSR process through the designed CS-SFT layers so that SR facial images obtain a promising balance between fidelity and realness. Unlike GAN inversion methods which necessitate costly image optimization at runtime, the proposed CFGPFSR can jointly recover facial details by only utilizing one forward pass. Experimental results on synthetic and real images indicate that the proposed CFGPFSR obtains remarkable performance in 16 × SR task, and some of its metrics such as peak signal to noise ratio (PSNR) and structural similarity (SSIM) are higher than that of the comparison methods. Meanwhile, it shows impressive results in reconstructing high-quality facial images.

Riesz Networks: Scale-Invariant Neural Networks in a Single Forward Pass

AbstractScale invariance of an algorithm refers to its ability to treat objects equally independently of their size. For neural networks, scale invariance is typically achieved by data augmentation. However, when presented with a scale far outside the range covered by the training set, neural networks may fail to generalize. Here, we introduce the Riesz network, a novel scale- invariant neural network. Instead of standard 2d or 3d convolutions for combining spatial information, the Riesz network is based on the Riesz transform which is a scale-equivariant operation. As a consequence, this network naturally generalizes to unseen or even arbitrary scales in a single forward pass. As an application example, we consider detecting and segmenting cracks in tomographic images of concrete. In this context, ‘scale’ refers to the crack thickness which may vary strongly even within the same sample. To prove its scale invariance, the Riesz network is trained on one fixed crack width. We then validate its performance in segmenting simulated and real tomographic images featuring a wide range of crack widths. An additional experiment is carried out on the MNIST Large Scale data set.

PReFormer: A memory-efficient transformer for point cloud semantic segmentation

The success of transformer networks in the natural language processing and 2D vision domains has encouraged the adaptation of transformers to 3D computer vision tasks. However, most of the existing approaches employ standard backpropagation (SBP). SBP requires the storage of model activations on a forward pass for use during the backward pass, making their memory complexity linearly proportional to model depth, hence, inefficient. Furthermore, most 3D point transformers use the classic QKV matrix multiplication design which comes with a memory bottleneck. To address these issues, we propose a memory-efficient point transformer that makes use of reversible functions and linearized self-attention to minimize SBP and transformer memory complexities, respectively. Additionally, rather than the usual UNet architectural design for segmentation, we adopt a ∇-shaped design to capture multi-size/resolution feature representations towards a finely detailed segmentation. Experimental results on benchmark datasets (Toronto3D, DALES, and CSPC) from different sensor platforms (vehicle, aerial, and backpack) show that our approach uses less than half the number of model parameters compared to its SBP counterpart. It also takes more than twice the input sequence and uses less than half the memory compared to most of the traditional approaches. Additionally, our use of a ∇-shaped architectural design yielded about a 5% increment in model performance over the U-shaped approach. Overall, the proposed PReFormer attained competitive performance compared to the state-of-the-art.

Employing neural density functionals to generate potential energy surfaces.

With the union of machine learning (ML) and quantum chemistry, amid the debate between machine-learned functionals and human-designed functionals in density functional theory (DFT), this paper aims to demonstrate the generation of potential energy surfaces using computations with machine-learned density functional approximation (ML-DFA). A recent research trend is the application of ML in quantum sciences in the design of density functionals such as DeepMind's Deep Learning model (DeepMind21, DM21). Though science reported the state-of-the-art performance of DM21, the opportunity to utilize DeepMind's pretrained DM21 neural networks in computations in quantum chemistry has not yet been tapped. So far in the literature, the Deep Learning density functionals (DM21) have not been applied to generate potential energy surfaces. While the superior accuracy of DM21 has been reported, there is still a scarcity of publications that apply DM21 in calculations in the field. In this context, for the first time in literature, neural density functionals inferring 2D potential energy surfaces (ML-DFA-PES) based on machine-learned DFA-based computational method is contributed in this paper. This paper reports the ML-DFA-generated PES for C4H8, H2O, H2, and H2+ by employing a pretrained DM21m TensorFlow model with cc-pVDZ basis set. In addition, we also analyze the long-range behavior of DM21 based PES to investigate the ability to describe a system at long ranges. Furthermore, we compare PES diagrams from DM21 with popular DFT functionals (b3lyp/ PW6B95) and CCSD(T). In this method, 2D potential energy surfaces are obtained using a method that relies upon the neural network's ability to accurately learn the mapping between 3D electron density and exchange-correlation potential. By inserting Deep Learning inference in DFT with a pretrained neural network, self-consistent field (SCF) energy at different geometries along the coordinates of interest is computed, and then, potential energy surfaces are plotted. In this method, first, the electron density is computed mathematically, and this computed 3D electron density is used as a ML feature vector to predict the exchange correlation potential as a ML inference computed by a forward pass of pre-trained DM21 TensorFlow computational graph, followed by the computation of self-consistent field energy at multiple geometries, and then, SCF energies at different bond lengths/angles are plotted as 2D PES. We implement this in a python source code using frameworks such as PySCF and DM21. This paper contributes this implementation in open source. The source code and DM21-DFA-based PES are contributed at https://sites.google.com/view/MLfunctionals-DeepMind-PES .

HSCNet++: Hierarchical Scene Coordinate Classification and Regression for Visual Localization with Transformer

Visual localization is critical to many applications in computer vision and robotics. To address single-image RGB localization, state-of-the-art feature-based methods match local descriptors between a query image and a pre-built 3D model. Recently, deep neural networks have been exploited to regress the mapping between raw pixels and 3D coordinates in the scene, and thus the matching is implicitly performed by the forward pass through the network. However, in a large and ambiguous environment, learning such a regression task directly can be difficult for a single network. In this work, we present a new hierarchical scene coordinate network to predict pixel scene coordinates in a coarse-to-fine manner from a single RGB image. The proposed method, which is an extension of HSCNet, allows us to train compact models which scale robustly to large environments. It sets a new state-of-the-art for single-image localization on the 7-Scenes, 12-Scenes, Cambridge Landmarks datasets, and the combined indoor scenes.

Binarizing by Classification: Is Soft Function Really Necessary?

Binary neural networks leverage Sign function to binarize weights and activations, which require gradient estimators to overcome its non-differentiability and will inevitably bring gradient errors during backpropagation. Although many hand-designed soft functions have been proposed as gradient estimators to better approximate gradients, their mechanism is not clear and there are still huge performance gaps between binary models and their full-precision counterparts. To address these issues and reduce gradient error, we propose to tackle network binarization as a binary classification problem and use a multi-layer perceptron (MLP) as the classifier in the forward pass and gradient estimator in the backward pass. Benefiting from the MLP’s theoretical capability to fit any continuous function, it can be adaptively learned to binarize networks and backpropagate gradients without any prior knowledge of soft functions. From this perspective, we further empirically justify that even a simple linear function can outperform previous complex soft functions. Extensive experiments demonstrate that the proposed method yields surprising performance both in image classification and human pose estimation tasks. Specifically, we achieve 65.7% top-1 accuracy of ResNet-34 on ImageNet dataset, with an absolute improvement of 2.6%. Moreover, we take binarization as a lightweighting approach for pose estimation models and propose well-designed binary pose estimation networks SBPN and BHRNet. When evaluating on the challenging Microsoft COCO keypoint dataset, the proposed method enables binary networks to achieve a mAP of up to 60.6 for the first time. Experiments conducted on real platforms demonstrate that BNN achieves a better balance between performance and computational complexity, especially when computational resources are extremely low.

Datum-Wise Inference in Structured Environments

In various application domains (e.g., health, psychology), experts use Bayesian networks to represent relationships among variables. However, these variables are not in practice directly observable, but can be instead inferred via noisy but costly features. Herein, we study the problem of datum–wise feature selection and classification in the case where the label of each data instance is described by a known Bayesian network, and features are available at a cost. The goal is to accurately classify each data instance, while keeping the feature acquisition cost minimum. To this end, we first propose a forward pass algorithm that sequentially acquires features to infer the label of each variable in the Bayesian network. During this process, the proposed algorithm uses both the acquired features and the Bayesian network relationships. In an effort to improve classification accuracy, we also devise a backward pass algorithm, which exploits Bayesian network relationships along with evidence. We discuss the computational complexity of both algorithms and experimentally assess their performance on 11 datasets. We observe that the forward pass algorithm achieves higher accuracy using a small fraction of features compared to state–of–the–art, while the backward pass algorithm enhances accuracy without acquiring additional features.

PC-JeDi: Diffusion for particle cloud generation in high energy physics

In this paper, we present a new method to efficiently generate jets in High Energy Physics called PC-JeDi. This method utilises score-based diffusion models in conjunction with transformers which are well suited to the task of generating jets as particle clouds due to their permutation equivariance. PC-JeDi achieves competitive performance with current state-of-the-art methods across several metrics that evaluate the quality of the generated jets. Although slower than other models, due to the large number of forward passes required by diffusion models, it is still substantially faster than traditional detailed simulation. Furthermore, PC-JeDi uses conditional generation to produce jets with a desired mass and transverse momentum for two different particles, top quarks and gluons.

Geochemical exploration and baseline from marine sediments in the fjords, channels and straits of Magellan, Chilean Patagonia

AbstractChilean Patagonia has a complex geological history dominated recently by Quaternary glaciers and rivers that have carried sediments from the mountains to depositional environments within marine fjords and channels. Although much is known about onshore placer deposits, little is known about these deposits and depositional systems in the offshore realm. This study analyzes the chemistry and mineralogy of marine sediments collected in the Strait of Magellan and around the region. This study aims to comprehend the distribution, concentration and provenance of critical/strategic elements and precious metals. In collaboration with the Cabo de Hornos research vessel of the Chilean Navy, 63 sediment samples were collected using a box core sampler, complemented with coastal beach samples. These samples underwent petrographic and geochemical analysis. The predominant opaque minerals identified in these samples are mainly iron oxides (magnetite–hematite‐ilmenite) and sulfides (pyrite, exceptional greigite). A comparison of the average element concentration for all samples for different baseline values revealed the following findings. As, Bi, Li, Re, Sb, Se, Te, and Pd exhibit positive anomalies for the Upper Continental Crust (UCC); Re, Se, and Sr. exhibit positive anomalies for Marine Sediments (MS), and Bi, Cs, Li, Re, Se, Sn, and Te show positive anomalies relative to the Folded Belt (FB). The main contrast anomalies occur in the following locations. Almirantazgo Sound (Sn, Ba, Sb, and Tm); Concepción Channel (Re, Sr., and Te); Smyth Channel (Mn, Re, Sb, Sr., Te, and Se); Almirante Montt Gulf (Mn, Re, and Sb); Nelson Strait (Re, Sr., and Te); Ancho Pass (Ti and Sr); Europa Fjord (Bi and Sn); Del Hambre Pass (Sr); Magdalena Channel and Sound (Sr); Froward Pass (Se), and Whiteside Channel (Mn). Regarding precious metals, silver exhibits contrasting anomalies relative to the UCC in seafloor sediments in the Almirante Montt Gulf (0.161 ppm) and Concepción Channel (0.101 ppm). Palladium contrast anomaly is observed in the Peel Estuary (8 ppb) and Platinum contrast anomalies occur in the Concepción (3 and 5 ppb) and Smyth Channels (4 ppb). Finally, gold, in comparison to the South Patagonian Batholith, reveals contrasting anomalies that occur in the Concepción Channel (8 ppb), Peel Estuary (6.7 and 8 ppb), Nelson Strait (7 ppb), Almirante Montt Gulf (6 ppb), Forward Pass (8 ppb), and Ancho Pass (7 ppb). In summary, the presented results provide valuable insights into the geological and geochemical baseline of the region, offering a better understanding of the relationship with depositional basins and source zones onshore during the late‐Quaternary.

Learning Diverse Tone Styles for Image Retouching.

Image retouching, aiming to regenerate the visually pleasing renditions of given images, is a subjective task where the users are with different aesthetic sensations. Most existing methods adopt a deterministic model to learn the retouching style from a specific expert, making it less flexible to meet diverse subjective preferences. Besides, the intrinsic diversity of an expert due to the targeted processing of different images is also deficiently described. To circumvent such issues, we propose to learn diverse image retouching with normalizing flow-based architectures. Unlike current flow-based methods which directly generate the output image, we argue that learning in a one-dimensional style space could 1) disentangle the retouching styles from the image content, 2) lead to a stable style presentation form, and 3) avoid the spatial disharmony effects. For obtaining meaningful image tone style representations, a joint-training pipeline is delicately designed, which is composed of a style encoder, a conditional RetouchNet, and the image tone style normalizing flow (TSFlow) module. In particular, the style encoder predicts the target style representation of an input image, which serves as the conditional information in the RetouchNet for retouching, while the TSFlow maps the style representation vector into a Gaussian distribution in the forward pass. After training, the TSFlow can generate diverse image tone style vectors by sampling from the Gaussian distribution. Extensive experiments on MIT-Adobe FiveK and PPR10K datasets show that our proposed method performs favorably against state-of-the-art methods and is effective in generating diverse results to satisfy different human aesthetic preferences. Source codeterministic and pre-trained models are publicly available at https://github.com/SSRHeart/TSFlow.

IMAGE RECONSTRUCTION OF OLD DAMAGED PHOTOS

In order to restore accurate and realistic details, blind face restoration often uses facial priors, such as a reference prior or a facial geometry prior. The applicability to real-world situations is, however, constrained by the inaccessibility of high quality references and the inability of very low-quality inputs to provide accurate geometric prior. In this paper, we present a GFP-GAN for blind face restoration that takes advantage of rich and varied priors included in a pre-trained face GAN. By the use of spatial feature transform layers, this Generative Facial Prior (GFP) is incorporated into the face restoration process, enabling our method to successfully strike a compromise between realism and fidelity. Whereas GAN inversion methods require image-specific tweaking at inference, our GFP-GAN could simultaneously restore facial details and enhance colours with just a single forward pass because of the powerful generative facial prior and delicate designs. Many tests demonstrate that, on both synthetic and realworld datasets, our technique outperforms earlier art.

Cyber vaccinator for image tamper resilient and recovery using invertible neural network

People frequently interact with their families, friends, and colleagues through Online Social Networks (OSNs). People post and share their photos in online communities and content-sharing sites. The problem addressed in this paper is the susceptibility of digital images to tampering, which compromises security and privacy. Traditional image forgery detection methods face challenges in reproducing original content after manipulation. This paper introduces an advanced Image Immunization System leveraging Invertible Neural Networks. The system, which comprises the cyber vaccinator, vaccine validator, forward pass for tamper detection, and backward pass for image self-recovery, aims to proactively immunize images against various attacks. The run-length encoding in the backward pass transforms hidden perturbations into information, facilitating the recovery of the authentic image. The middleware's expansion to multimodal content analysis, including videos and audio, provides a more comprehensive defense against digital manipulation within OSNs. These advancements reflect a commitment to robust security and holistic content integrity. The Cyber Vaccinator, using Invertible Neural Networks (INNs) for image tamper resilience and recovery, demonstrates significant effectiveness in detecting tampering and restoring images, providing a robust solution for maintaining image integrity. The Cyber Vaccinator uses an Invertible Neural Network (INN) to safeguard image integrity. It detects tampering by analyzing invariant features and responds with precise recovery methods. By continuously monitoring images, it ensures real-time tamper detection and efficient restoration, maintaining image authenticity through advanced neural network resilience and recovery techniques.

Physics to the Rescue: Deep Non-Line-of-Sight Reconstruction for High-Speed Imaging.

Computational approach to imaging around the corner, or non-line-of-sight (NLOS) imaging, is becoming a reality thanks to major advances in imaging hardware and reconstruction algorithms. A recent development towards practical NLOS imaging, Nam et al. [1] demonstrated a high-speed non-confocal imaging system that operates at 5Hz, 100x faster than the prior art. This enormous gain in acquisition rate, however, necessitates numerous approximations in light transport, breaking many existing NLOS reconstruction methods that assume an idealized image formation model. To bridge the gap, we present a novel deep model that incorporates the complementary physics priors of wave propagation and volume rendering into a neural network for high-quality and robust NLOS reconstruction. This orchestrated design regularizes the solution space by relaxing the image formation model, resulting in a deep model that generalizes well on real captures despite being exclusively trained on synthetic data. Further, we devise a unified learning framework that enables our model to be flexibly trained using diverse supervision signals, including target intensity images or even raw NLOS transient measurements. Once trained, our model renders both intensity and depth images at inference time in a single forward pass, capable of processing more than 5 captures per second on a high-end GPU. Through extensive qualitative and quantitative experiments, we show that our method outperforms prior physics and learning based approaches on both synthetic and real measurements. We anticipate that our method along with the fast capturing system will accelerate future development of NLOS imaging for real world applications that require high-speed imaging.

Enhancing adversarial robustness with randomized interlayer processing

Adversarial training, coupled with loss regularization techniques (such as MART and TRADES), is the current most effective method for consistently achieving adversarial robustness on various known datasets. However, without extra training data, the robustness gains of adversarial training are limited. To overcome this limitation, several alternative defenses were proposed as potential extensions to adversarial training, the most notable being the recent denoising diffusion models to generate additional training data. In this work, we propose a different candidate defense that combines both adversarial retraining and input transformation techniques, but instead, applies the transformations between the architecture layers without the need for extra training data. Namely, our interlayer processing technique introduces bit-depth reduction, originally an input pre-processing technique, between the layers of the model to reduce the space for an adversary to exploit. Together with adversarial training and randomization in the forward pass, interlayer processing leads to higher robustness gains. Our experiments show that our defense improves over standard loss regularization techniques by 1.56% and 7.92% for the ResNet-18 model on the CIFAR-10 and SVHN datasets, respectively. Additional experiments on ResNet-34 architecture led to improvements of 1.96% and 10.42% on CIFAR-100 and SVHN, respectively.

Channel-Agnostic Training of Transmitter and Receiver for Wireless Communications.

Wireless communications systems are traditionally designed by independently optimising signal processing functions based on a mathematical model. Deep learning-enabled communications have demonstrated end-to-end design by jointly optimising all components with respect to the communications environment. In the end-to-end approach, an assumed channel model is necessary to support training of the transmitter and receiver. This limitation has motivated recent work on over-the-air training to explore disjoint training for the transmitter and receiver without an assumed channel. These methods approximate the channel through a generative adversarial model or perform gradient approximation through reinforcement learning or similar methods. However, the generative adversarial model adds complexity by requiring an additional discriminator during training, while reinforcement learning methods require multiple forward passes to approximate the gradient and are sensitive to high variance in the error signal. A third, collaborative agent-based approach relies on an echo protocol to conduct training without channel assumptions. However, the coordination between agents increases the complexity and channel usage during training. In this article, we propose a simpler approach for disjoint training in which a local receiver model approximates the remote receiver model and is used to train the local transmitter. This simplified approach performs well under several different channel conditions, has equivalent performance to end-to-end training, and is well suited to adaptation to changing channel environments.

Learning the Geodesic Embedding with Graph Neural Networks

We present GEGNN, a learning-based method for computing the approximate geodesic distance between two arbitrary points on discrete polyhedra surfaces with constant time complexity after fast precomputation. Previous relevant methods either focus on computing the geodesic distance between a single source and all destinations, which has linear complexity at least or require a long precomputation time. Our key idea is to train a graph neural network to embed an input mesh into a high-dimensional embedding space and compute the geodesic distance between a pair of points using the corresponding embedding vectors and a lightweight decoding function. To facilitate the learning of the embedding, we propose novel graph convolution and graph pooling modules that incorporate local geodesic information and are verified to be much more effective than previous designs. After training, our method requires only one forward pass of the network per mesh as precomputation. Then, we can compute the geodesic distance between a pair of points using our decoding function, which requires only several matrix multiplications and can be massively parallelized on GPUs. We verify the efficiency and effectiveness of our method on ShapeNet and demonstrate that our method is faster than existing methods by orders of magnitude while achieving comparable or better accuracy. Additionally, our method exhibits robustness on noisy and incomplete meshes and strong generalization ability on out-of-distribution meshes. The code and pretrained model can be found on https://github.com/IntelligentGeometry/GeGnn.

UPL-SFDA: Uncertainty-Aware Pseudo Label Guided Source-Free Domain Adaptation for Medical Image Segmentation.

Domain Adaptation (DA) is important for deep learning-based medical image segmentation models to deal with testing images from a new target domain. As the source-domain data are usually unavailable when a trained model is deployed at a new center, Source-Free Domain Adaptation (SFDA) is appealing for data and annotation-efficient adaptation to the target domain. However, existing SFDA methods have a limited performance due to lack of sufficient supervision with source-domain images unavailable and target-domain images unlabeled. We propose a novel Uncertainty-aware Pseudo Label guided (UPL) SFDA method for medical image segmentation. Specifically, we propose Target Domain Growing (TDG) to enhance the diversity of predictions in the target domain by duplicating the pre-trained model's prediction head multiple times with perturbations. The different predictions in these duplicated heads are used to obtain pseudo labels for unlabeled target-domain images and their uncertainty to identify reliable pseudo labels. We also propose a Twice Forward pass Supervision (TFS) strategy that uses reliable pseudo labels obtained in one forward pass to supervise predictions in the next forward pass. The adaptation is further regularized by a mean prediction-based entropy minimization term that encourages confident and consistent results in different prediction heads. UPL-SFDA was validated with a multi-site heart MRI segmentation dataset, a cross-modality fetal brain segmentation dataset, and a 3D fetal tissue segmentation dataset. It improved the average Dice by 5.54, 5.01 and 6.89 percentage points for the three tasks compared with the baseline, respectively, and outperformed several state-of-the-art SFDA methods.

Path-Constrained Time-Optimal Motion Planning for Robot Manipulators With Third-Order Constraints

Planning time-optimal motions of robot manipulators is important for maximizing productivity in industry and has attracted much attention in academia. Time-optimal time scaling (TOTS) of specified paths subject to second-order constraints is a well-studied problem in robotics. However, there is still a lack of effective approaches for computing TOTS with higher order constraints. This article proposes a novel method for TOTS subject to third-order constraints. The proposed method formulates the third-order TOTS as a four-stage optimization involving only linear programming by exploiting the special structure of the problem. The proposed method first performs a backward pass and a forward pass to generate the second-order optimal velocity profile. Then, the points violating the third-order constraints are identified, and the constraint violations are eliminated through numerical integration around the violating points. A bisection search algorithm is proposed to quickly determine the switching points at which the elimination process begins. The proposed method is verified through numerical examples and experimental tests. The results indicate that the proposed method outperforms the state-of-the-art approaches in terms of solution quality and computation time.

Two‐stage diffusion model deriving FDG‐PET from T1 Weighted Magnetic Resonance Images for diagnosis of Alzheimer’s disease

AbstractBackgroundAlzheimer’s Disease (AD) is a neurodegenerative disease which is characterized by cognitive and functional impairment, and defined by triad of Amyloid status, Tau pathology, and Neuroimaging marker of neurodegeneration or neuronal injury. An early diagnosis of AD can be challenging and it can be detected earlier in brain functional imaging than brain structural imaging.MethodWe develop a pipeline for synthesizing FDG‐PET/MRI from source MRI‐T1‐Weighted (MRI‐T1W) images based on denoising diffusion mechanism [1]. The pipeline includes a forward pass which encodes the source MRI‐T1WI images to gaussian noise space, and a backward pass which generates the corresponding FDG‐PET from the derived noises with repeated denoising steps. Furthermore, to validate the effectiveness of the synthesized FDG‐PET images, we also develop multiple types of classifiers and the experiments for both single input of MRI‐T1WI and joint input of MRI‐T1WI and synthesized FDG‐PET.ResultThe experiments are performed on 1036 of FDG‐PET/MRI pairs (286,474, 276 for Cognitive Normal, Mild Cognitive Impaiement and Alzheimer’s Disease each) are obtained from ADNI [2] datasets and display the promising results with low Mean Absolute Error (MAE) and high Structural Similarity Index Measure (SSIM) for the two stage diffusion denoising based FDG‐PET synthesis pipeline. The results for the additional classification experiments also show that in most cases the joint input enhances the classifiers' performance in accuracy and F1‐score. The results also indicate that the synthesized FDG‐PET images are not only visually similar to the ground truth images, but also improve the performance of machine learning based AD diagnosis.ConclusionThis work provides support for the notion that machine learning‐derived image analysis may be a useful approach to improving the diagnosis of AD.Reference:[1] Ho, J., Jain, A., & Abbeel, P. (2020). Denoising diffusion probabilistic models. Advances in Neural Information Processing Systems, 33, 6840‐6851.[2] Petersen, R., Aisen, P. S., Beckett, L. A., Donohue, M. C., Gamst, A. C., Harvey, D. J., … & Toga, A. W. (2010). Alzheimer’s disease neuroimaging initiative (ADNI). Neurology, 74(3), 201‐209.