Retraining deep neural networks when new data arrives is typically computationally expensive. Moreover, certain applications do not allow such costly retraining due to time or computational constraints. Fast Bayesian updates are a possible solution to this issue. Therefore, we propose a Bayesian update based on Monte-Carlo samples and a last-layer Laplace approximation for different Bayesian neural network types, i.e., Dropout, Ensemble, and Spectral Normalized Neural Gaussian Process (SNGP). In a large-scale evaluation study, we show that our updates combined with SNGP represent a fast and competitive alternative to costly retraining. As a use case, we combine the Bayesian updates for SNGP with different sequential query strategies to exemplarily demonstrate their improved selection performance in active learning.
This paper describes a high-fidelity sonar model and a simulation environment that implements the model. The model and simulation environment have been developed to aid in the design of a forward looking sonar for autonomous underwater vehicles (AUVs). The simulator achieves real-time visualization through ray tracing and approximation. The simulator facilitates the assessment of sonar design choices, such as beam pattern and beam location, and assessment of obstacle detection and tracking algorithms. An obstacle detection model is proposed for which the null hypothesis is estimated from the environmental model. Sonar data is generated from the simulator and compared to the expected results from the detection model demonstrating the benefits and limitations of the proposed approach.
The majority of real-world processes are spatiotemporal, and the data generated by them exhibits both spatial and temporal evolution. Weather is one of the most important processes that fall under this domain, and forecasting it has become a crucial part of our daily routine. Weather data analysis is considered the most complex and challenging task. Although numerical weather prediction models are currently state-of-the-art, they are resource intensive and time-consuming. Numerous studies have proposed time-series-based models as a viable alternative to numerical forecasts. Recent research has primarily focused on forecasting weather at a specific location. Therefore, models can only capture temporal correlations. This self-contained paper explores various methods for regional data-driven weather forecasting, i.e., forecasting over multiple latitude-longitude points to capture spatiotemporal correlations. The results showed that spatiotemporal prediction models reduced computational cost while improving accuracy; in particular, the proposed tensor train dynamic mode decomposition-based forecasting model has comparable accuracy to ConvLSTM without the need for training. We use the NASA POWER meteorological dataset to evaluate the models and compare them with the current state of the art.
BERT is a widely used pre-trained model in natural language processing. However, because its time and space requirements increase with a quadratic level of the text length, the BERT model is difficult to use directly on the long-text corpus. The collected text data is usually quite long in some fields, such as health care. Therefore, to apply the pre-trained language knowledge of BERT to long text, in this paper, imitating the skimming-intensive reading method used by humans when reading a long paragraph, the Skimming-Intensive Model (SkIn) is proposed. It can dynamically select the critical information in the text so that the length of the input into the BERT-Base model is significantly reduced, which can effectively save the cost of the classification algorithm. Experiments show that the SkIn method has achieved better results than the baselines on long-text classification datasets in the medical field, while its time and space requirements increase linearly with the text length, alleviating the time and space overflow problem of BERT on long-text data.
MD-GAN is a machine learning-based method that can evolve part of the system at any time step, accelerating the generation of molecular dynamics data. For the accurate prediction of MD-GAN, sufficient information on the dynamics of a part of the system should be included with the training data. Therefore, the selection of the part of the system is important for efficient learning. In a previous study, only one particle (or vector) of each molecule was extracted as part of the system. Therefore, we investigated the effectiveness of adding information from other particles to the learning process. In the experiment of the polyethylene system, when the dynamics of three particles of each molecule were used, the diffusion was successfully predicted using one-third of the time length of the training data, compared to the single-particle input. Surprisingly, the unobserved transition of diffusion in the training data was also predicted using this method.
RGBD-based real-time dynamic 3D reconstruction suffers from inaccurate inter-frame motion estimation as errors may accumulate with online tracking. This problem is even more severe for single-view-based systems due to strong occlusions. Based on these observations, we propose OcclusionFusion, a novel method to calculate occlusion-aware 3D motion to guide the reconstruction. In our technique, the motion of visible regions is first estimated and combined with temporal information to infer the motion of the occluded regions through an LSTM-involved graph neural network. Furthermore, our method computes the confidence of the estimated motion by modeling the network output with a probabilistic model, which alleviates untrustworthy motions and enables robust tracking. Experimental results on public datasets and our own recorded data show that our technique outperforms existing single-view-based real-time methods by a large margin. With the reduction of the motion errors, the proposed technique can handle long and challenging motion sequences. Please check out the project page for sequence results: https://wenbin-lin.github.io/OcclusionFusion.
We introduce a spherical fingertip sensor for dynamic manipulation. It is based on barometric pressure and time-of-flight proximity sensors and is low-latency, compact, and physically robust. The sensor uses a trained neural network to estimate the contact location and three-axis contact forces based on data from the pressure sensors, which are embedded within the sensor's sphere of polyurethane rubber. The time-of-flight sensors face in three different outward directions, and an integrated microcontroller samples each of the individual sensors at up to 200 Hz. To quantify the effect of system latency on dynamic manipulation performance, we develop and analyze a metric called the collision impulse ratio and characterize the end-to-end latency of our new sensor. We also present experimental demonstrations with the sensor, including measuring contact transitions, performing coarse mapping, maintaining a contact force with a moving object, and reacting to avoid collisions.
Deep Neural Networks ( DNN s) are known to make overconfident mistakes, which makes their use problematic in safety-critical applications. State-of-the-art ( SOTA ) calibration techniques improve on the confidence of predicted labels alone and leave the confidence of non-max classes (e.g. top-2, top-5) uncalibrated. Such calibration is not suitable for label refinement using post-processing. Further, most SOTA techniques learn a few hyper-parameters post-hoc, leaving out the scope for image, or pixel specific calibration. This makes them unsuitable for calibration under domain shift, or for dense prediction tasks like semantic segmentation. In this paper, we argue for intervening at the train time itself, so as to directly produce calibrated DNN models. We propose a novel auxiliary loss function: Multi-class Difference in Confidence and Accuracy ( MDCA ), to achieve the same MDCA can be used in conjunction with other application/task-specific loss functions. We show that training with MDCA leads to better-calibrated models in terms of Expected Calibration Error ( ECE ), and Static Calibration Error ( SCE ) on image classification, and segmentation tasks. We report ECE ( SCE ) score of 0.72 (1.60) on the CIFAR 100 dataset, in comparison to 1.90 (1.71) by the SOTA. Under domain shift, a ResNet-18 model trained on PACS dataset using MDCA gives an average ECE ( SCE ) score of 19.7 (9.7) across all domains, compared to 24.2 (11.8) by the SOTA. For the segmentation task, we report a 2X reduction in calibration error on PASCAL - VOC dataset in comparison to Focal Loss. Finally, MDCA training improves calibration even on imbalanced data, and for natural language classification tasks. We have released the code here: code is available at https://github.com/mdca-loss
In this paper, we tackle the problem of domain shift. Most existing methods perform training on multiple source domains using a single model, and the same trained model is used on all unseen target domains. Such solutions are sub-optimal as each target domain exhibits its own speciality, which is not adapted. Furthermore, expecting the single-model training to learn extensive knowledge from the multiple source domains is counterintuitive. The model is more biased toward learning only domain-invariant features and may result in negative knowledge transfer. In this work, we propose a novel framework for unsupervised test-time adaptation, which is formulated as a knowledge distillation process to address domain shift. Specifically, we incorporate Mixture-of-Experts (MoE) as teachers, where each expert is separately trained on different source domains to maximize their speciality. Given a test-time target domain, a small set of unlabeled data is sampled to query the knowledge from MoE. As the source domains are correlated to the target domains, a transformer-based aggregator then combines the domain knowledge by examining the interconnection among them. The output is treated as a supervision signal to adapt a student prediction network toward the target domain. We further employ meta-learning to enforce the aggregator to distill positive knowledge and the student network to achieve fast adaptation. Extensive experiments demonstrate that the proposed method outperforms the state-of-the-art and validates the effectiveness of each proposed component. Our code is available at https://github.com/n3il666/Meta-DMoE.
Roboy 3.0 is an open-source tendon-driven humanoid robot that mimics the musculoskeletal system of the human body. Roboy 3.0 is being developed as a remote robotic body - or a robotic avatar - for humans to achieve remote physical presence. Artificial muscles and tendons allow it to closely resemble human morphology with 3-DoF neck, shoulders and wrists. Roboy 3.0 3-DoF joints are implemented as ball-and-socket joints. While industry provides a clear solution for 1-DoF joint pose sensing, it is not the case for the ball-and-socket joint type. In this paper we present a custom solution to estimate the pose of a ball-and-socket joint. We embed an array of magnets into the ball and an array of 3D magnetic sensors into the socket. We then, based on the changes in the magnetic field as the joint rotates, are able to estimate the orientation of the joint. We evaluate the performance of two neural network approaches using the LSTM and Bayesian-filter like DVBF. Results show that in order to achieve the same mean square error (MSE) DVBFs require significantly more time training and hyperparameter tuning compared to LSTMs, while DVBF cope with sensor noise better. Both methods are capable of real-time joint pose estimation at 37 Hz with MSE of around 0.03 rad for all three degrees of freedom combined. The LSTM model is deployed and used for joint pose estimation of Roboy 3.0's shoulder and neck joints. The software implementation and PCB designs are open-sourced under https://github.com/Roboy/ball_and_socket_estimator