Continual configuration tuning in cellular radio access networks (RANs) is critical for maintaining performance, reliability, energy efficiency, and user experience. However, this task remains largely manual in practice. Automating it needs to confront high-dimensional configuration spaces, sparse and biased exploration, strong parameter interactions, and substantial environmental confounding. Existing RAN configuration tuning approaches have limited effectiveness in addressing these challenges. In this paper, we present CausalTune, a novel causal learning framework for automated RAN configuration optimization based on observational telemetry. CausalTune disentangles configuration effects from environmental and operational confounders, generalizes to sparse and previously unseen parameter settings, and captures high-impact multi-parameter interactions. Our key insight is that effective causal inference in operational RANs requires reshaping raw telemetry to expose confounding and learning environment-invariant mechanisms. Guided by this insight, CausalTune employs a multistage pipeline that integrates distributional representation learning, causal modeling, and interaction-aware recommendation. We evaluate CausalTune using 10 months of RAN measurement data from 1M commercial cells of a major cellular operator. Our comparison to state-of-the-art baselines shows that CausalTune achieves up to 3X KPI improvement on the held-out dataset. In terms of causal modeling quality, CausalTune achieves up to 12X lower KPI reconstruction error; on recommended configuration safety, it achieves 4X higher agreement with expert engineers while significantly reducing off-target recommendations. These findings demonstrate the potential of causal learning to enable reliable, scalable, and interpretable RAN configuration tuning.

Three-Dimensional (3D) semantic segmentation is an essential building block for interactive Augmented Reality (AR). However, existing Deep Neural Network (DNN) models for segmenting 3D objects are not only computation-intensive but also memory heavy, hindering their deployment on resource-constrained mobile devices. We present the design, implementation and evaluation of Slimmer, a generic and model-independent framework for accelerating 3D semantic segmentation and facilitating its real-time applications on mobile devices. In contrast to the current practice that directly feeds a point cloud to DNN models, Slimmer is motivated by our observation that these models remain high accuracy even if we remove a fraction of points from the input, which can significantly reduce the inference time and memory usage of these models. Our design of Slimmer faces two key challenges. First, the simplification method of point clouds should be lightweight. Otherwise, the reduced inference time may be canceled out by the incurred overhead of input-data simplification. Second, Slimmer still needs to accurately segment the removed points from the input to create a complete segmentation of the original input, again, using a lightweight method. Our extensive performance evaluation demonstrates that, by addressing these two challenges, Slimmer can dramatically reduce the resource utilization of a representative DNN model for 3D semantic segmentation. For example, if we can tolerate 1% accuracy loss, the reduction could be ~20% for inference time and ~9% for memory usage. The reduction increases to around ~27% for inference time and ~15% for memory usage when we can tolerate 2% accuracy loss.

Visual exploration of volumetric datasets to discover the embedded features and spatial structures is a challenging and tedious task. In this paper we present a semi-automatic approach to this problem that works by visually segmenting the intensity-gradient 2D histogram of a volumetric dataset into an exploration hierarchy. Our approach mimics user exploration behavior by analyzing the histogram with the normalized-cut multilevel segmentation technique. Unlike previous work in this area, our technique segments the histogram into a reasonable set of intuitive components that are mutually exclusive and collectively exhaustive. We use information-theoretic measures of the volumetric data segments to guide the exploration. This provides a data-driven coarse-to-fine hierarchy for a user to interactively navigate the volume in a meaningful manner.

The field of visualization has addressed navigation of very large datasets, usually meshes and volumes. Significantly less attention has been devoted to the issues surrounding navigation of very large images. In the last few years the explosive growth in the resolution of camera sensors and robotic image acquisition techniques has widened the gap between the display and image resolutions to three orders of magnitude or more. This paper presents the first steps towards navigation of very large images, particularly landscape images, from an interactive visualization perspective. The grand challenge in navigation of very large images is identifying regions of potential interest. In this paper we outline a three-step approach. In the first step we use multi-scale saliency to narrow down the potential areas of interest. In the second step we outline a method based on statistical signatures to further cull out regions of high conformity. In the final step we allow a user to interactively identify the exceptional regions of high interest that merit further attention. We show that our approach of progressive elicitation is fast and allows rapid identification of regions of interest. Unlike previous work in this area, our approach is scalable and computationally reasonable on very large images. We validate the results of our approach by comparing them to user-tagged regions of interest on several very large landscape images from the Internet.

While benchmark datasets have been proposed for testing computer vision and 3D shape retrieval algorithms, no such datasets have yet been put forward to assess the relevance of these techniques for engineering problems. This paper presents several distinctive benchmark datasets for evaluating techniques for automated classification and retrieval of CAD objects. These datasets include (1) a dataset of CAD primitives (such as those common in constructive solid geometry modeling); (2) two datasets consisting of classes generated by minor topological variation; (3) two datasets of industrial CAD models classified based on object function and manufacturing process, respectively; (4) and a dataset of LEGO© models from the Mindstorms© robotics kits. Each model in the datasets is available in three formats - ACIS SAT, ISO STEP, and as a VRML mesh (some models are available under several different fidelity settings). These are all available through the National Design Repository.Using these datasets, we present comprehensive empirical results for nińe (9) different shape and solid model matching and retrieval techniques. These experiments show, as expected, that the quality of precision-recall performance can significantly vary on different datasets. These experiments reveal that for certain object classes and classifications, such as those based on manufacturing processes, all existing techniques perform poorly. This study reveals the strengths and weaknesses of existing research in these areas, introduces open challenge problems, and provides meaningful datasets and metrics against which the success of current and future work can be measured.

This paper used surface curvatures and support vector machines for the identification of manufacturing processes in a database of mesh based CAD artifacts. The target is to classify prismatic machined and cast-then-machined parts into their respective classification to assist the manufacturing cost estimation. Current research on shape matching techniques has used shape functions to match the gross shapes of mesh models. However, they do not adequately discriminate artifacts manufactured by different processes. Our approach shows how different kinds of surfaces can distinguish different manufacturing processes. Statistics on surface curvatures are used to construct shape descriptors; then supervised machine learning classifier support vector machines are applied to separate the two classifications.