Temporal Smoothing Research Articles

BundleMoCap++: Efficient, robust and smooth motion capture from sparse multiview videos

Producing smooth and accurate motions from sparse videos without requiring specialized equipment and markers is a long-standing problem in the research community. Most approaches typically involve complex processes such as temporal constraints, multiple stages combining data-driven regression and optimization techniques, and bundle solving over temporal windows. These increase the computational burden and introduce the challenge of hyperparameter tuning for the different objective terms. In contrast, BundleMoCap++ offers a simple yet effective approach to this problem. It solves the motion in a single stage, eliminating the need for temporal smoothness objectives while still delivering smooth motions without compromising accuracy. BundleMoCap++ outperforms the state-of-the-art without increasing complexity. Our approach is based on manifold interpolation between latent keyframes. By relying on a local manifold smoothness assumption and appropriate interpolation schemes, we efficiently solve a bundle of frames using two or more latent codes. Additionally, the method is implemented as a sliding window optimization and requires only the first frame to be properly initialized, reducing the overall computational burden. BundleMoCap++’s strength lies in achieving high-quality motion capture results with fewer computational resources. To do this efficiently, we propose a novel human pose prior that focuses on the geometric aspect of the latent space, modeling it as a hypersphere, allowing for the introduction of sophisticated interpolation techniques. We also propose an algorithm for optimizing the latent variables directly on the learned manifold, improving convergence and performance. Finally, we introduce high-order interpolation techniques adapted for the hypersphere, allowing us to increase the solving temporal window, enhancing performance and efficiency.

A shaping two-stage anomaly data recovery method based on multi-norm joint optimization under energy internet

With the deep integration of energy networks and information communication technology, the involvement of hardware and environmental factors intensifies the challenges to data security. Prior research focuses predominantly on interpolating random missing data under the assumption of no data damage, without providing strategies for addressing continuous data loss. To overcome, we proposes a novel two-stage data recovery method, that includes three-fold ideas: (1) it introduces the matrix shaping differentiation strategy to realize the differentiated recovery of random and continuous missing data; (2) it designs a two-stage data recovery approach, in which the first stage constructs a random missing data recovery objective function by eliminating the consecutive missing data columns, and the second stage performs temporal pre-interpolation backfilling on the eliminated columns and introduces the temporal correlation to construct the temporal smoothing objective function; and (3) it develops an alternating direction method of multipliers-based iterative solution algorithm. Experimental results on four public datasets demonstrate that our model is superior to five state-of-the-art and classical methods in terms of random missing error rate, random column missing error rate, root mean square error, and mean absolute error.

Missing data recovery based on temporal smoothness and time-varying similarity for wireless sensor network

Wireless Sensor Networks (WSN) play a vital role in the Internet of Things (IoT) and show great potential in monitoring applications. However, due to harsh environmental conditions and unreliable communication links, WSN often encounter partial data loss during data collection, which inevitably affects the quality of service. To address this challenge, researchers have employed matrix completion techniques to recover missing data by exploiting the low-rank features in the data, but its accuracy is not satisfactory. This paper argues that the spatiotemporal characteristics of the data underlie its low-rank nature, enabling a more accurate capture of the intrinsic patterns within the data. Drawing on this insight, we propose a missing data recovery algorithm based on Temporal Smoothness and Time-Varying Similarity (TSTVS). Unlike traditional low-rank methods, the TSTVS algorithm directly utilizes the structural features of data in the spatiotemporal domain to establish a missing data completion model. Subsequently, the model is converted into an unconstrained optimization problem using the penalty function method, and the gradient descent method is applied to solve it, reconstructing the complete data matrix. Finally, simulation experiments were conducted on three real-world monitoring datasets, comparing the TSTVS with three low-rank methods, Efficient Data Collection Approach (EDCA), Matrix factorization with Smoothness constraints (MFS) and Data Recovery Based on Low Rank and Short-Term Stability(DRLRSS). The experimental results indicate that the proposed TSTVS algorithm consistently outperforms the three low-rank based algorithms in terms of recovery accuracy across different datasets and missing rate scenarios.

Adaptive Gaussian Markov random fields for child mortality estimation.

The under-5 mortality rate (U5MR), a critical health indicator, is typically estimated from household surveys in lower and middle income countries. Spatio-temporal disaggregation of household survey data can lead to highly variable estimates of U5MR, necessitating the usage of smoothing models which borrow information across space and time. The assumptions of common smoothing models may be unrealistic when certain time periods or regions are expected to have shocks in mortality relative to their neighbors, which can lead to oversmoothing of U5MR estimates. In this paper, we develop a spatial and temporal smoothing approach based on Gaussian Markov random field models which incorporate knowledge of these expected shocks in mortality. We demonstrate the potential for these models to improve upon alternatives not incorporating knowledge of expected shocks in a simulation study. We apply these models to estimate U5MR in Rwanda at the national level from 1985 to 2019, a time period which includes the Rwandan civil war and genocide.

Dynamic undirected graphical models for time-varying clinical symptom and neuroimaging networks.

In this work, we propose methods to examine how the complex interrelationships between clinical symptoms and, separately, brain imaging biomarkers change over time leading up to the diagnosis of a disease in subjects with a known genetic near-certainty of disease. We propose a time-dependent undirected graphical model that ensures temporal and structural smoothness across time-specific networks to examine the trajectories of interactions between markers aligned at the time of disease onset. Specifically, we anchor subjects relative to the time of disease diagnosis (anchoring time) as in a revival process, and we estimate networks at each time point of interest relative to the anchoring time. To use all available data, we apply kernel weights to borrow information across observations that are close to the time of interest. Adaptive lasso weights are introduced to encourage temporal smoothness in edge strength, while a novel elastic fused- penalty removes spurious edges and encourages temporal smoothness in network structure. Our approach can handle practical complications such as unbalanced visit times. We conduct simulation studies to compare our approach with existing methods. We then apply our method to data from PREDICT-HD, a large prospective observational study of pre-manifest Huntington's disease (HD) patients, to identify symptom and imaging network changes that precede clinical diagnosis of HD.

A moisture budget perspective on Australian rainfall variability

AbstractRainfall variability over Australia is revisited from the viewpoint of the atmospheric moisture budgets in three regions: the extratropics, Subtropics, and Tropics. The budgets are calculated using three‐hourly European Centre for Medium‐Range Weather Forecasts Reanalysis v5 (ERA5) and ERA5‐Land data between 1979 and 2022. The use of the moisture budget at short time‐scales enables the investigation of the relationship between synoptic weather‐scale processes and the longer term variability of the rainfall climate. The total variability in the vertically integrated moisture flux divergence (VIMD) is significantly larger than the evaporation minus precipitation (E − P), to a large extent due to the sub‐daily time‐scales. E − P is related more closely to moisture flux convergence in winter (summer) over south (north) Australia, suggesting a clear seasonality in the relationship between the two budget terms. The E − P–VIMD relationship is nearly in phase in the Tropics, whereas VIMD leads E − P by 9–15 hr with eastward‐propagating signals in the extratropics and Subtropics. Such seasonal and regional discrepancies in the relationship are attributed to the background state of moisture availability and temperature as represented by relative humidity and lifting condensation levels. The variability of the budget imbalance and its seasonality are dominated by the variability in VIMD. The imbalance reduces rapidly with temporal smoothing, with the storage term approaching zero at approximately 20 days, which can be thought of as making a transition time‐scale from high‐frequency weather‐related variability into slow‐varying background conditions. Weather‐related variability (cyclones, fronts, and thunderstorms) dominates the overall E − P variability in the extratropics and Subtropics, whereas slow‐varying background conditions contribute equally to the total variability in the Tropics.

Incorporation of a modified temporal cepstrum smoothing in both signal-to-noise ratio and speech presence probability estimation for speech enhancement

Incorporation of a modified temporal cepstrum smoothing in both signal-to-noise ratio and speech presence probability estimation for speech enhancement

Deformation-encoding Deep Learning Transformer for High-Frame-Rate Cardiac Cine MRI.

Purpose To develop a deep learning model for increasing cardiac cine frame rate while maintaining spatial resolution and scan time. Materials and Methods A transformer-based model was trained and tested on a retrospective sample of cine images from 5840 patients (mean age, 55 years ± 19 [SD]; 3527 male patients) referred for clinical cardiac MRI from 2003 to 2021 at nine centers; images were acquired using 1.5- and 3-T scanners from three vendors. Data from three centers were used for training and testing (4:1 ratio). The remaining data were used for external testing. Cines with downsampled frame rates were restored using linear, bicubic, and model-based interpolation. The root mean square error between interpolated and original cine images was modeled using ordinary least squares regression. In a prospective study of 49 participants referred for clinical cardiac MRI (mean age, 56 years ± 13; 25 male participants) and 12 healthy participants (mean age, 51 years ± 16; eight male participants), the model was applied to cines acquired at 25 frames per second (fps), thereby doubling the frame rate, and these interpolated cines were compared with actual 50-fps cines. The preference of two readers based on perceived temporal smoothness and image quality was evaluated using a noninferiority margin of 10%. Results The model generated artifact-free interpolated images. Ordinary least squares regression analysis accounting for vendor and field strength showed lower error (P < .001) with model-based interpolation compared with linear and bicubic interpolation in internal and external test sets. The highest proportion of reader choices was "no preference" (84 of 122) between actual and interpolated 50-fps cines. The 90% CI for the difference between reader proportions favoring collected (15 of 122) and interpolated (23 of 122) high-frame-rate cines was -0.01 to 0.14, indicating noninferiority. Conclusion A transformer-based deep learning model increased cardiac cine frame rates while preserving both spatial resolution and scan time, resulting in images with quality comparable to that of images obtained at actual high frame rates. Keywords: Functional MRI, Heart, Cardiac, Deep Learning, High Frame Rate Supplemental material is available for this article. © RSNA, 2024.

Measurement properties of movement smoothness metrics for upper limb reaching movements in people with moderate to severe subacute stroke

BackgroundMovement smoothness is a potential kinematic biomarker of upper extremity (UE) movement quality and recovery after stroke; however, the measurement properties of available smoothness metrics have been poorly assessed in this group. We aimed to measure the reliability, responsiveness and construct validity of several smoothness metrics.MethodsThis ancillary study of the REM-AVC trial included 31 participants with hemiparesis in the subacute phase of stroke (median time since stroke: 38 days). Assessments performed at inclusion (Day 0, D0) and at the end of a rehabilitation program (Day 30, D30) included the UE Fugl Meyer Assessment (UE-FMA), the Action Research Arm Test (ARAT), and 3D motion analysis of the UE during three reach-to-point movements at a self-selected speed to a target located in front at shoulder height and at 90% of arm length. Four smoothness metrics were computed: a frequency domain smoothness metric, spectral arc length metric (SPARC); and three temporal domain smoothness metrics (TDSM): log dimensionless jerk (LDLJ); number of submovements (nSUB); and normalized average rectified jerk (NARJ).ResultsAt D30, large clinical and kinematic improvements were observed. Only SPARC and LDLJ had an excellent reliability (intra-class correlation > 0.9) and a low measurement error (coefficient of variation < 10%). SPARC was responsive to changes in movement straightness (rSpearman=0.64) and to a lesser extent to changes in movement duration (rSpearman=0.51) while TDSM were very responsive to changes in movement duration (rSpearman>0.8) and not to changes in movement straightness (non-significant correlations). Most construct validity hypotheses tested were verified except for TDSM with low correlations with clinical metrics at D0 (rSpearman<0.5), ensuing low predictive validity with clinical metrics at D30 (non-significant correlations).ConclusionsResponsiveness and construct validity of TDSM were hindered by movement duration and/or noise-sensitivity. Based on the present results and concordant literature, we recommend using SPARC rather than TDSM in reaching movements of uncontrolled duration in individuals with spastic paresis after stroke.Trial RegistrationNCT01383512, https://clinicaltrials.gov/, June 27, 2011.

Meshless track assimilation (MTA) of 3D PTV data

We propose a data assimilation algorithm to improve the resolution of 3D particle tracking velocimetry (PTV) data and, as a byproduct, it estimates the pressure field. The proposed method, called meshless tracking assimilation (MTA), does not rely on a mesh to compute derivatives. Instead, it exploits the analytical formulation of the obtained field to extract the gradients of the involved quantities in arbitrary points. MTA actively constraints PTV data points to satisfy the Navier–Stokes equations in both time and space at user-defined spatial locations. Three test cases are investigated. The first one is a synthetic dataset reconstructed from a direct numerical simulation and it serves to ensure that the method is properly implemented and working. The second one is the 2020 data assimilation challenge which gives a qualitative comparison with other data assimilation methods. Finally, MTA is used on a Tomographic PTV experiment of a Jet Flow with ReD≃3000 to ensure it can be implemented successfully on experimental data. The first and third test cases show a significant performance boost compared to the classic linear interpolation. This improvement allows for a higher level of structural resolution without sacrificing the temporal and spatial smoothness of the solution. The second test case generally aligns well with other state-of-the-art data assimilation techniques in terms of performance, but there are some noticeable differences.

Feasibility of Estimating Sea Surface Height Anomalies from Surface Ocean Currents and Winds

Abstract We propose a method to reconstruct sea surface height anomalies (SSHA) from vector surface currents and winds. This analysis is motivated by the proposed satellite ODYSEA, which is a Doppler scatterometer that measures coincident surface vector winds and currents. If it is feasible to estimate SSHA from these measurements, then ODYSEA could provide collocated fields of SSHA, currents, and winds over a projected wide swath of ∼1700 km. The reconstruction also yields estimates of the low-frequency surface geostrophic, Ekman, irrotational, and nondivergent current components and a framework for separation of balanced and unbalanced motions. The reconstruction is based on a steady-state surface momentum budget including the Ekman drift, Coriolis acceleration, and horizontal advection. The horizontal SSHA gradient is obtained as a residual of these terms, and the unknown SSHA is solved for using a Helmholtz–Hodge decomposition given an imposed SSHA boundary condition. We develop the reconstruction using surface currents, winds, and SSHA off the U.S. West Coast from a 43-day coupled ROMS–WRF simulation. We also consider how simulated ODYSEA measurement and sampling errors and boundary condition uncertainties impact reconstruction accuracy. We find that temporal smoothing of the currents for periods of 150 h is necessary to mitigate large reconstruction errors associated with unbalanced near-inertial motions. For the most realistic case of projected ODYSEA measurement noise and temporal sampling, the reconstructed SSHA fields have an RMS error of 2.1 cm and a model skill (squared correlation) of 0.958 with 150-h resolution. We conclude that an accurate SSHA reconstruction is feasible using information measured by ODYSEA and external SSHA boundary conditions.

Improved Landsat Operational Land Imager (OLI) Cloud and Shadow Detection with the Learning Attention Network Algorithm (LANA)

Landsat cloud and cloud shadow detection has a long heritage based on the application of empirical spectral tests to single image pixels, including the Landsat product Fmask algorithm, which uses spectral tests applied to optical and thermal bands to detect clouds and uses the sun-sensor-cloud geometry to detect shadows. Since the Fmask was developed, convolutional neural network (CNN) algorithms, and in particular U-Net algorithms (a type of CNN with a U-shaped network structure), have been developed and are applied to pixels in square patches to take advantage of both spatial and spectral information. The purpose of this study was to develop and assess a new U-Net algorithm that classifies Landsat 8/9 Operational Land Imager (OLI) pixels with higher accuracy than the Fmask algorithm. The algorithm, termed the Learning Attention Network Algorithm (LANA), is a form of U-Net but with an additional attention mechanism (a type of network structure) that, unlike conventional U-Net, uses more spatial pixel information across each image patch. The LANA was trained using 16,861 512 × 512 30 m pixel annotated Landsat 8 OLI patches extracted from 27 images and 69 image subsets that are publicly available and have been used by others for cloud mask algorithm development and assessment. The annotated data were manually refined to improve the annotation and were supplemented with another four annotated images selected to include clear, completely cloudy, and developed land images. The LANA classifies image pixels as either clear, thin cloud, cloud, or cloud shadow. To evaluate the classification accuracy, five annotated Landsat 8 OLI images (composed of &gt;205 million 30 m pixels) were classified, and the results compared with the Fmask and a publicly available U-Net model (U-Net Wieland). The LANA had a 78% overall classification accuracy considering cloud, thin cloud, cloud shadow, and clear classes. As the LANA, Fmask, and U-Net Wieland algorithms have different class legends, their classification results were harmonized to the same three common classes: cloud, cloud shadow, and clear. Considering these three classes, the LANA had the highest (89%) overall accuracy, followed by Fmask (86%), and then U-Net Wieland (85%). The LANA had the highest F1-scores for cloud (0.92), cloud shadow (0.57), and clear (0.89), and the other two algorithms had lower F1-scores, particularly for cloud (Fmask 0.90, U-Net Wieland 0.88) and cloud shadow (Fmask 0.45, U-Net Wieland 0.52). In addition, a time-series evaluation was undertaken to examine the prevalence of undetected clouds and cloud shadows (i.e., omission errors). The band-specific temporal smoothness index (TSIλ) was applied to a year of Landsat 8 OLI surface reflectance observations after discarding pixel observations labelled as cloud or cloud shadow. This was undertaken independently at each gridded pixel location in four 5000 × 5000 30 m pixel Landsat analysis-ready data (ARD) tiles. The TSIλ results broadly reflected the classification accuracy results and indicated that the LANA had the smallest cloud and cloud shadow omission errors, whereas the Fmask had the greatest cloud omission error and the second greatest cloud shadow omission error. Detailed visual examination, true color image examples and classification results are included and confirm these findings. The TSIλ results also highlight the need for algorithm developers to undertake product quality assessment in addition to accuracy assessment. The LANA model, training and evaluation data, and application codes are publicly available for other researchers.

Towards Understanding Future: Consistency Guided Probabilistic Modeling for Action Anticipation

Action anticipation aims to infer the action in the unobserved segment (future segment) with the observed segment (past segment). Existing methods focus on learning key past semantics to predict the future, but they do not model the temporal continuity between the past and the future. However, past actions are always highly uncertain in anticipating the unobserved future. The absence of temporal continuity smoothing in the video's past-and-future segments may result in an inconsistent anticipation of future action. In this work, we aim to smooth the global semantics changes in the past and future segments. We propose a Consistency-guided Probabilistic Model (CPM), which focuses on learning the globally temporal probabilistic consistency to inhibit the unexpected temporal consistency. The CPM is deployed on the Transformer architecture, which includes three modules of future semantics estimation, global semantics estimation, and global distribution estimation involving the learning of past-to-future semantics, past-and-future semantics, and semantically probabilistic distributions. To achieve the smoothness of temporal continuity, we follow the principle of variational analysis and describe two probabilistic distributions, i.e., a past-aware distribution and a global-aware distribution, which help to estimate the evidence lower bound of future anticipation. In this study, we maximize the evidence lower bound of future semantics by reducing the distribution distance between the above two distributions for model optimization. Extensive experiments demonstrate that the effectiveness of our method and the CPM achieves state-of-the-art performance on Epic-Kitchen100, Epic-Kitchen55, and EGTEA-GAZE.

Spear and Shield: Adversarial Attacks and Defense Methods for Model-Based Link Prediction on Continuous-Time Dynamic Graphs

Real-world graphs are dynamic, constantly evolving with new interactions, such as financial transactions in financial networks. Temporal Graph Neural Networks (TGNNs) have been developed to effectively capture the evolving patterns in dynamic graphs. While these models have demonstrated their superiority, being widely adopted in various important fields, their vulnerabilities against adversarial attacks remain largely unexplored. In this paper, we propose T-SPEAR, a simple and effective adversarial attack method for link prediction on continuous-time dynamic graphs, focusing on investigating the vulnerabilities of TGNNs. Specifically, before the training procedure of a victim model, which is a TGNN for link prediction, we inject edge perturbations to the data that are unnoticeable in terms of the four constraints we propose, and yet effective enough to cause malfunction of the victim model. Moreover, we propose a robust training approach T-SHIELD to mitigate the impact of adversarial attacks. By using edge filtering and enforcing temporal smoothness to node embeddings, we enhance the robustness of the victim model. Our experimental study shows that T-SPEAR significantly degrades the victim model's performance on link prediction tasks, and even more, our attacks are transferable to other TGNNs, which differ from the victim model assumed by the attacker. Moreover, we demonstrate that T-SHIELD effectively filters out adversarial edges and exhibits robustness against adversarial attacks, surpassing the link prediction performance of the naive TGNN by up to 11.2% under T-SPEAR. The code and datasets are available at https://github.com/wooner49/T-spear-shield

Complex expressional characterizations learning based on block decomposition for temporal knowledge graph completion

A temporal knowledge graph (TKG) is a set of facts associated with different timestamps. TKG completion (TKGC) is the task of inferring unknown facts based on known facts, where mining and understanding the expressional characterization of timestamps from the fact sets are key. Expressional characterizations are complex because of two aspects: type-diversity and dependence-variability. Existing approaches do not comprehensively consider time points and time intervals; hence, facts with different time forms cannot be managed simultaneously. Additionally, these approaches overlook the interdependence and variability of timestamp positions in different types of facts, such as the relative order (e.g., whether the facts belong to the past, present, or future) and relative distance (e.g., the largest distance at which the facts occur), thus hindering the adequate perception of position. Therefore, we propose a novel TKGC model named Complex Expressional Characterizations with Block Decomposition (CEC-BD). This model is based on the block decomposition model and exploits sophisticated interactions for different types of facts by unifying the treatment of both time points and time intervals. In addition, we introduce a time-sensitive approach that injects timestamps with embedded information regarding fact occurrences into the CEC-BD model to capture the relative order and distance. To cluster and order time embeddings more effectively, we add a bias component that is randomly initialized and learned during training and further propose a temporal smoothness scheme. In particular, the model achieved a mean reciprocal rank of 3.05% and 10 times the speedup of state-of-the-art baseline models.

Learning to Remove Rain in Video With Self-Supervision.

In heavy rain video, rain streak and rain accumulation are the most common causes of degradation. They occlude background information and can significantly impair the visibility. Most existing methods rely heavily on the synthetic training data, and thus raise the domain gap problem that prevents the trained models from performing adequately in real testing cases. Unlike these methods, we introduce a self-learning method to remove both rain streaks and rain accumulation without using any ground-truth clean images in training our model, which consequently can alleviate the domain gap issue. The main idea is based on the assumptions that (1) adjacent clean frames can be aligned or warped from one frame to another frame, (2) rain streaks are distributed randomly in the temporal domain, (3) the rain streak/accumulation related variables/priors can be inferred reliably from the information within the images/sequences. Based on these assumptions, we construct an augmented Self-Learned Deraining Network (SLDNet+) to remove both rain streaks and rain accumulation by utilizing temporal correlation, consistency, and rain-related priors. For the temporal correlation, our SLDNet+ takes rain degraded adjacent frames as its input, aligns them, and learns to predict the clean version of the current frame. For the temporal consistency, a new loss is designed to build a robust mapping between the predicted clean frame and non-rain regions from the adjacent rain frames. For the rain-streak-related prior, the rain streak removal network is optimized jointly with motion estimation and rain region detection; while for the rain-accumulation-related prior, a novel non-local video rain accumulation removal method is developed to estimate the accumulation-lines from the whole input video and to offer better color constancy and temporal smoothness. Extensive experiments show the effectiveness of our approach, which provides superior results compared with the existing state of the art methods both quantitatively and qualitatively. The source code will be made publicly available at: https://github.com/flyywh/CVPR-2020-Self-Rain-Removal-Journal.

Anomaly Detection in CCTV Surveillance

CCTV surveillance systems are routinely utilized to maintain the safety and security of public and private locations. However, manually watching surveillance footage may be tiresome and time-consuming, making it difficult to recognize and respond to possible threats quickly. In this research, we offer a real-time danger detection system for CCTV monitoring that uses deep learning models to identify and categorize degrees of high movement in video frames. Our system is able to continuously monitor surveillance footage in real-time and identify potential threats such as abuse, burglaries, explosions, shootings, fighting, shoplifting, road accidents, arson, robbery, stealing, assault, and vandalism by treating videos as segments and defining anomalous (threatening) and normal (safe) segments. We conducted extensive tests on a huge collection of CCTV video to evaluate the effectiveness of our system and obtained encouraging findings. Our solution has the potential to greatly increase the efficiency and efficacy of CCTV surveillance, allowing for faster reaction times and improved individual security. We use multiple instance learning (MIL) to automatically develop a deep anomaly ranking model that predicts high anomaly scores for anomalous video segments by treating normal and anomalous films as bags and video segments as instances. In addition, we apply sparsity and temporal smoothness requirements in the ranking loss function to improve anomaly localization during training. In addition, we present a novel large-scale, first-of-its-kind dataset comprising 128 hours of video. It comprises of 1900 uncut real-world surveillance movies with 13 actual anomalies such as fights, car accidents, burglary, robbery, and so on, as well as typical activities. This dataset may be used for two different purposes. First, global anomaly detection, which takes into account all abnormalities in one group and all normal activity in another. Secondly, for identifying every one of the thirteen unusual actions. Comparing our experimental results to state-of-the-art methodologies, we find that our MIL method for anomaly detection produces a considerable increase on anomaly detection performance. We provide the findings from many recent deep learning baselines concerning the identification of aberrant behavior. These baselines' poor recognition performance indicates how difficult our dataset is, and it also provides additional room for investigation in the future.

Manifold Regularizer for High-Resolution fMRI Joint Reconstruction and Dynamic Quantification.

Oscillating Steady-State Imaging (OSSI) is a recently developed fMRI acquisition method that can provide 2 to 3 times higher SNR than standard fMRI approaches. However, because the OSSI signal exhibits a nonlinear oscillation pattern, one must acquire and combine nc (e.g., 10) OSSI images to get an image that is free of oscillation for fMRI, and fully sampled acquisitions would compromise temporal resolution. To improve temporal resolution and accurately model the nonlinearity of OSSI signals, instead of using subspace models that are not well suited for the data, we build the MR physics for OSSI signal generation as a regularizer for the undersampled reconstruction. Our proposed physics-based manifold model turns the disadvantages of OSSI acquisition into advantages and enables joint reconstruction and quantification. OSSI manifold model (OSSIMM) outperforms subspace models and reconstructs high-resolution fMRI images with a factor of 12 acceleration and without spatial or temporal smoothing. Furthermore, OSSIMM can dynamically quantify important physics parameters, including R2∗ maps, with a temporal resolution of 150 ms.

Validation of single reference variable flip angle (SR-VFA) dynamic T1 mapping with T2 * correction using a novel rotating phantom.

To validate single reference variable flip angle (SR-VFA) dynamic T1 mapping with and without T2 * correction against inversion recovery (IR) T1 measurements. A custom cylindrical phantom with three concentric compartments was filled with variably doped agar to produce a smooth spatial gradient of the T1 relaxation rate as a function of angle across each compartment. IR T1 , VFA T1 , and B1 + measurements were made on the phantom before rotation, and multi-echo stack-of-radial dynamic images were acquired during rotation via an MRI-compatible motor. B1 + -corrected SR-VFA and SR-VFA-T2 * T1 maps were computed from the sliding window reconstructed images and compared against rotationally registered IR and VFA T1 maps to determine the percentage error. Both VFA and SR-VFA-T2 * T1 maps fell within 10% of IR T1 measurements for a low rotational speed, with a mean accuracy of 2.3% ± 2.6% and 2.8% ± 2.6%, respectively. Increasing rotational speed was found to decrease the accuracy due to increasing temporal smoothing over ranges where the T1 change had a nonconstant slope. SR-VFA T1 mapping was found to have similar accuracy as the SR-VFA-T2 * and VFA methods at low TEs (˜<2 ms), whereas accuracy degraded strongly with later TEs. T2 * correction of the SR-VFA T1 maps was found to consistently improve accuracy and precision, especially at later TEs. SR-VFA-T2 * dynamic T1 mapping was found to be accurate against reference IR T1 measurements within 10% in an agar phantom. Further validation is needed in mixed fat-water phantoms and in vivo.

Spatio-temporal smoothing and dynamics of different electricity flexibility options for highly renewable energy systems—Case study for Norway

In this article, we investigate the mismatch of renewable electricity production to demand and how flexibility options enabling spatial and temporal smoothing can reduce risks of variability. As a case study we pick a simplified (partial) 2-region representation of the Norwegian electricity system and focus on wind power. We represent regional electricity production and demand through two stochastic processes: the wind capacity factors are modelled as a two-dimensional Ornstein–Uhlenbeck process and electricity demand consists of realistic base load and temperature-induced load coming from a deseasonalised autoregressive process. We validate these processes, that we have trained on historical data, through Monte Carlo simulations allowing us to generate many statistically representative weather years. For the investigated realisations (weather years) we study deviations of production from demand under different wind capacities, and introduce different scenarios where flexibility options like storage and transmission are available. Our analysis shows that simulated loss values are reduced significantly by cooperation between regions and either mode of flexibility. Combining storage and transmission leads to even more synergies and helps to stabilise production levels and thus reduces likelihoods of inadequacy of renewable power systems.