Learning Actions That Reduce Variation in Objects

Many works have posited the benefit of depth in deep networks. However, one of the problems encountered in the training of very deep networks is feature reuse; that is, features are ‘diluted’ as they are forward propagated through the model. Hence, later network layers receive less informative signals about the input data, consequently making training less effective. In this work, we address the problem of feature reuse by taking inspiration from an earlier work which employed residual learning for alleviating the problem of feature reuse. We propose a modification of residual learning for training very deep networks to realize improved generalization performance; for this, we allow stochastic shortcut connections of identity mappings from the input to hidden layers. We perform extensive experiments using the USPS and MNIST datasets. On the USPS dataset, we achieve an error rate of 2.69% without employing any form of data augmentation (or manipulation). On the MNIST dataset, we reach a comparable state-of-the-art error rate of 0.52%. Particularly, these results are achieved without employing any explicit regularization technique.

Visualizing similarity data of different objects by exhibiting more separate organizations with local and multimodal characteristics preserved is important in multivariate data analysis. Laplacian Eigenmaps (LAE) and Locally Linear Embedding (LLE) aim at preserving the embeddings of all similarity pairs in the close vicinity of the reduced output space, but they are unable to identify and separate interclass neighbors. This paper considers the semi-supervised manifold learning problems. We apply the pairwise Cannot-Link and Must-Link constraints induced by the neighborhood graph to specify the types of neighboring pairs. More flexible regulation on supervised information is provided. Two novel multimodal nonlinear techniques, which we call trace ratio (TR) criterion-based semi-supervised LAE (S <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> LAE) and LLE (S <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> LLE), are then proposed for marginal manifold visualization. We also present the kernelized S <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> LAE and S <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> LLE. We verify the feasibility of S <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> LAE and S <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> LLE through extensive simulations over benchmark real-world MIT CBCL, CMU PIE, MNIST, and USPS data sets. Manifold visualizations show that S <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> LAE and S <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> LLE are able to deliver large margins between different clusters or classes with multimodal distributions preserved. Clustering evaluations show they can achieve comparable to or even better results than some widely used methods.

A cumulative descriptor enhanced ensemble deep neural networks method for remaining useful life prediction of cutting tools

Using ambient assisted living infrastructures to monitor emotional change: Applications in supporting dementia interventions

In feature learning field, many methods are inspired by advances in neuroscience. Among them, neural network and sparse coding have been broadly studied. Predictive sparse decomposition (PSD) is a practical variant of these two methods. It trains a neural network to estimate the sparse codes. After training, the neural network is fine-tuned to achieve higher performance on object recognition tasks. It is widely believed that introducing discriminative information can make the features more useful for classification task. Hence, in this work, we propose applying the task-driven dictionary learning framework to the PSD and demonstrate that this new model can be optimized by the stochastic gradient descent (SGD) algorithm. Before our work, the semi-supervised auto-encoder framework has already been proposed to guide neural network to extract discriminative representations. But it does not improve the classification performance of neural network. In the experiments, we compare the proposed method with the semi-supervised auto-encoder method. The performance of PSD is used as the baseline for these two methods. On the MNIST and USPS datasets, our method can generate more discriminative and predictable sparse codes than other methods. Furthermore, the recognition accuracy of neural network can be improved.

Semi-supervised clustering can be viewed as a clustering paradigm that exploits both labeled and unlabeled data to steer learning accurate data clusters and avoid local minimum solutions. Nonetheless, the attempts to refine existing semi-supervised clustering methods are relatively limited when compared to the advancements witnessed in the current benchmark methods in fully unsupervised clustering. This research introduces a novel semi-supervised method for deep clustering that leverages deep neural networks and fuzzy memberships to better capture the data partitions. In particular, the proposed Dual-Constraint-based Semi-Supervised Deep Clustering (DC-SSDEC) method utilizes two sets of pairwise soft constraints; "should-link" and "shouldNot-link", to guide the clustering process. The intended clustering task is expressed as an optimization of a newly designed objective function. Additionally, DC-SSDEC performance was evaluated through comprehensive experiments using three real-world and benchmark datasets. Moreover, a comparison with related state-of-the-art clustering techniques was conducted to showcase the DC-SSDEC outperformance. In particular, DC-SSDEC significance consists of the proposed dual-constraint formulation and its integration into a novel objective function. This contribution yielded an improvement in the resulting clustering performance compared to relevant state-of-the-art approaches. In addition, the assessment of the proposed model using real-world datasets represents another contribution of this research. In fact, increases of 3.25%, 1.44%, and 1.82% in the clustering accuracy were gained by DC-SSDEC over the best performing single-constraint-based approach, using MNIST, STL-10, and USPS datasets, respectively.

Understanding the intricate relationship between sensory perception and physicochemical properties of cacao-based products is crucial for advancing quality control and driving product innovation. However, effectively integrating these heterogeneous data sources poses a significant challenge, particularly when sensory evaluations are derived from low-quality, subjective, and often inconsistent annotations provided by multiple experts. We propose a comprehensive framework that leverages a correlated chained Gaussian processes model for learning from crowds, termed MAR-CCGP, specifically designed for a customized Casa Luker database that integrates sensory and physicochemical data on cacao-based products. By formulating sensory evaluations as regression tasks, our approach enables the estimation of continuous perceptual scores from physicochemical inputs, while concurrently inferring the latent, input-dependent reliability of each annotator. To address the inherent noise, subjectivity, and non-stationarity in expert-generated sensory data, we introduce a three-stage methodology: (i) construction of an integrated database that unifies physicochemical parameters with corresponding sensory descriptors; (ii) application of a MAR-CCGP model to infer the underlying ground truth from noisy, crowd-sourced, and non-stationary sensory annotations; and (iii) development of a novel localized expert trustworthiness approach, also based on MAR-CCGP, which dynamically adjusts for variations in annotator consistency across the input space. Our approach provides a robust, interpretable, and scalable solution for learning from heterogeneous and noisy sensory data, establishing a principled foundation for advancing data-driven sensory analysis and product optimization in the food science domain. We validate the effectiveness of our method through a series of experiments on both semi-synthetic data and a novel real-world dataset developed in collaboration with Casa Luker, which integrates sensory evaluations with detailed physicochemical profiles of cacao-based products. Compared to state-of-the-art learning-from-crowds baselines, our framework consistently achieves superior predictive performance and more precise annotator reliability estimation, demonstrating its efficacy in multi-annotator regression settings. Of note, our unique combination of a novel database, robust noisy-data regression, and input-dependent trust scoring sets MAR-CCGP apart from existing approaches.

The Web becomes a large knowledge database in the world. Most of the data stored in the Web are text based. Recently, many kinds of sensors, including obstacle sensors, video cameras, thermometers, and so on, have been installed in various places in our daily living environments: stations, streets, malls, etc. It becomes important to open the data obtained by the sensors existing in our daily living environment for various purposes. Since the data obtained by observing the real world directly with sensors include real-world information different from the Web, a new worldwide social information infrastructure, which we call Sensing Web, may be realized. In the workshop, we would like to discuss technical research issues related to pattern recognition for realizing the Sensing Web. The fields of interest are as follows, but not limited: • Sensor networks, sensor data transmission • Sensor data processing, including visual data, audio data, access logs, and other sensor data • Privacy handling for sensor data • Sensor data description for data exchange and data retrieval • Sensor data presentation • Other applications for sensor data

CubeSats continue to proliferate and are an excellent low-cost method of remote sensing. A key piece of intelligent systems is sensory input, data storage, and data communications. With the continued miniaturization of technology, CubeSats will increase their sensory inputs with future miniaturization and enhance their robustness for autonomous operations if data and communications are secure. These futures inspire an intelligent system solution to on-orbit communications. This paper explores a dual-microprocessor approach to improve hardware cybersecurity of intelligent systems, with a view toward intensional intelligence as a means of adjudicating access to sensitive data onboard the CubeSat. With enhanced cybersecurity, Artificial Intelligent Systems (AIS) will add vital utility to otherwise vulnerable, autonomous systems. Using Systems Models-Based Thinking, we shed light on our plan to apply artificial intelligent system concepts to advance CubeSat technology. Managing technology for AIS reduces some of the uncertainties and risks associated with the space environment.

This paper presents a modern control system for three phase high performance interior permanent magnet motor drives. The control system combines the advantages of vector control and direct torque control and overcomes some of the implementation drawbacks of either of the methods. The system uses a current control loop in connection with a switching table. A brain emotional learning based intelligent controller (BELBIC) is employed in the system which provides excellent performance under different operating conditions. BELBIC is based on the emotion processing mechanism in human's brain providing actions based on sensory inputs and emotional cues. A comprehensive evaluation of an IPM motor performance under the proposed control system proves the effectiveness of the system and its partial superiority over either vector control or direct torque control despite its relatively simple structure.

Hierarchical clustering is one of the most suitable tools to discover the underlying true structure of a dataset in the case of unsupervised learning where the ground truth is unknown and classical machine learning classifiers are not suitable. In many real applications, it provides a perspective on inner data structure and is preferred to partitional methods. However, determining the resulting number of clusters in hierarchical clustering requires human expertise to deduce this from the dendrogram and this represents a major challenge in making a fully automatic system such as the ones required for decision support in Industry 4.0. This research proposes a general criterion to perform the cut of a dendrogram automatically, by comparing six original criteria based on the Calinski-Harabasz index. The performance of each criterion on 95 real-life dendrograms of different topologies is evaluated against the number of classes proposed by the experts and a winner criterion is determined. This research is framed in a bigger project to build an Intelligent Decision Support system to assess the performance of 3D printers based on sensor data in real-time, although the proposed criteria can be used in other real applications of hierarchical clustering.The methodology is applied to a real-life dataset from the 3D printers and the huge reduction in CPU time is also shown by comparing the CPU time before and after this modification of the entire clustering method. It also reduces the dependability on human-expert to provide the number of clusters by inspecting the dendrogram. Further, such a process allows applying hierarchical clustering in an automatic mode in real-life industrial applications and allows the continuous monitoring of real 3D printers in production, and helps in building an Intelligent Decision Support System to detect operational modes, anomalies, and other behavioral patterns.

Ubiquitous mobile sensors on human activity recognition pose the threat of leaking personal information that is implicitly contained within the time-series sensor signals and can be extracted by attackers. Existing protective methods only support specific sensitive attributes and require massive relevant sensitive ground truth for training, which is unfavourable to users. To fill this gap, we propose a novel data transformation framework for prohibiting the leakage of sensitive information from sensor data. The proposed framework transforms raw sensor data into a new format, where the sensitive information is hidden and the desired information (e.g., human activities) is retained. Training can be conducted without using any personal information as ground truth. Meanwhile, multiple attributes of sensitive information (e.g., age, gender) can be collectively hidden through a one-time transformation. The experimental results on two multimodal sensor-based human activity datasets manifest the feasibility of the presented framework in hiding users’ sensitive information (inference MAE increases <inline-formula><tex-math notation="LaTeX">$\sim$</tex-math></inline-formula> 2 times and inference accuracy degrades <inline-formula><tex-math notation="LaTeX">$\sim$</tex-math></inline-formula> 50%) without degrading the usability of the data for activity recognition (only <inline-formula><tex-math notation="LaTeX">$\sim$</tex-math></inline-formula> 2% accuracy degradation).

A valuable asset for the improvement of aviation maintenance is the correct assessment of the aircraft systems health condition, for a more accurate planning and execution of maintenance routines. As such, the creation of a Prognostic and Health Management (PHM) system, supported by Condition Based Maintenance (CBM) can bring important benefits to the aeronautical field. The ultimate goal of a PHM system is the correct prediction of the Remaining Useful Lifetime (RUL) of a certain aircraft system, using the sensors data extracted during flights. Nevertheless, a relevant stage in the PHM pipeline concerns the estimation of the system condition, expressed by the Health Indicator (HI). The HI value reflects the health condition of a specific aircraft system, which can possibly be affected by degradation, failures or external conditions occurred during flight time. Henceforth, due to the relevancy of the HI assessment for the accuracy of the PHM model, this paper aims to propose a new formulation for the HI computation, derived from raw anonymized data retrieved from different sensors within the aircraft system. The proposed formulation combines information from the different variables (like sensors) that have an impact on the overall system condition, by assigning a positive or negative weight to each variable depending on the influence on the system behaviour. The weights are determined based on the typical and standard data patterns. Thus, the estimated HI aims to reflect the number of hours of flight expected to be flown, based only on raw data extracted from the system. Furthermore, considering that the available sensors data is anonymized, a study of the relevancy of the different sensors features for the degradation assessment is performed, using specific metrics. Considering a scenario where the HI ground truth is unknown, the failure data of each aircraft system is used to evaluate and discuss the formulation suitability. The HI formulation is applied in real datasets, on the environmental systems of two wide body aircraft types.

Human and AI collaboration has evolved drastically over the past decade. Most of the tasks are now partly handled by machines and partly by humans, resulting in human and AI partnership. Humans share their tasks with the AI agents as they are good at performing predictive and repetitive tasks. The tasks assigned to the agents can be dynamic in nature and can change over time. This may lead to acquiring new skills or re-training to improvise on the same task. The developer needs to programme these agents every time for a new domain. In this work, we propose a framework that helps in developing apprentice agents that learn from Human demonstration. Apprenticeship learning is the process of learning by observing an expert performing actions to achieve a specific goal. The agent observes the expert's actions, the sensory inputs and then trains itself using the observed data. We also propose a process to learn and self-diagnose the agent's action for apprenticeship learning. The framework alleviates the need to explicitly program the apprentice agents. We demonstrate our framework on a case study.

The 2023 Mw 6.9 Al Haouz Earthquake in the Moroccan High Atlas mountains is the most recent example of a destructive event in an intraplate setting under the absence of a causative plate boundary. Their large inter-event times in such seismotectonic regimes hinders the study of single focus regions and apparently underestimates their larger seismic hazard potential, leaving local communities unaware of the scale of their exposure to these risks. One major shortcoming in the immediate scientific analysis of this unprecedented earthquake is the lack of (publicly) available seismic waveform recordings in the source area.An immediate impact is displayed in the source location uncertainty, including a high variability in focal depth estimations. Associating the earthquake origin to a causative fault remains puzzling, bearing in mind the deeper-than-average focal depth of around 32 km. The complex tectonic history of the Atlas mountain chains is depicted in a plethora of active, reactivated and abandoned faults. In addition, prior studies noticed a lithospheric anomaly underneath the High Atlas mountain range, that lacks a classical mountain root. Thus, this event raise questions on the thermo-elastic parameters at depth in order to support a large seismogenic thickness of the crust.In the immediate aftermath of the Al Haouz, we initiated a rapid-response task force for setting up a temporal seismic network in the High Atlas. Within two weeks, six autonomous seismological stations have been installed between 20 and 80 km from the estimated epicenter. Due to the large degree of destruction, problematic access and absence of major infrastructure in the epicentral region, we relied on autarkic and robust sensors. As a novelty, the stations consisted of SmartSolo&amp;#174; three component 5 Hz geophone sensors. These industrial-level instruments are light-weight and easy to deploy in any terrain. In a compact casing, the passive sensors host digitizer, data storage, GPS and battery life-time of up to 30 days.Here, we present the first data of this 2 months temporary network. Based on preliminary analysis, we could obtain more than 1000 events recorded on all stations of the network during the 54 recording days. The largest recorded events had an assigned magnitude of ML 4.5. Considering single stations detections as well, we estimate more than 10,000 earthquake detections overall. This catalog will greatly expand the scientific insight into the mechanisms of this exceptional earthquake in the near-future.Further, through the use of simple geophone sensors this study presents a proof of concept for rapid response installations of temporary aftershock network. These sensor types strongly outperform the fully elaborated counterparts in installation speed, transportability and external requirements for potential installation sites without large drawbacks in data quality.