Closest Competitor Research Articles

The financial and industrial situation of NVIDIA

NVIDIA has been gaining the world’s attention in recent years. Since its operation, it has shown great creativity and productivity. Most people in the world will come up with NVIDIA in mind when it comes to gaming GPUs. Thus, a brief introduction to this company seems significant. Previous studies have looked into technologies, investments and some specific ratios. However, few focus mainly on financial statements and the industrial environment. This passage gives insight into the income statements and balance sheets from the 2022-2024 fiscal years of NVIDIA, together with an environmental SWOT analysis. After these studies, NVIDIA has been stated to have a robust financial situation and a stable position among close competitors. NVIDIA owns the industry environment and a larger seller market. The passage also shows some of the strategies NVIDIA has adopted or is suitable for it to adopt. Further studies may be included afterwards in a more empirical way.

EXPRESS: Retailer Differentiation in Social Media: an Investigation of Firm-Generated Content on Twitter

Social media platforms have been used by firms for a variety of purposes - for building firms’ brand image, increasing customer engagement, providing customer service, among others. However, there is very little research on content strategies adopted by traditional rival firms competing on online social media platforms. This paper seeks to fill this gap by examining whether retailers, traditionally identified as close competitors, mirror this rivalry in their social media content strategies on Twitter. To this end, this study introduces a new metric for assessing competition on online social media, based on content similarity. The authors find that retailers competing closely in traditional context show greater divergence in their content strategies on social media, and firms whose social media content strategies are less similar to content strategies of their close traditional rivals benefit from higher engagement and acquire new followers faster. In examining the mechanism of the effect, the authors find that these divergent firms’ improved performance is attributable to their superior ability to leverage the higher-level affordances of Twitter as compared to their rivals. The results of this study offer valuable insights for firms seeking to distinguish their social media content from that of their competitors.

Integration of Generative-Adversarial-Network-Based Data Compaction and Spatial Attention Transductive Long Short-Term Memory for Improved Rainfall–Runoff Modeling

This work presents a novel approach to rainfall–runoff modeling. We incorporate GAN-based data compaction into a spatial-attention-enhanced transductive long short-term memory (TLSTM) network. The GAN component reduces data dimensions while retaining essential features. This compaction enables the TLSTM to capture complex temporal dependencies in rainfall–runoff patterns more effectively. When tested on the CAMELS dataset, the model significantly outperforms benchmark LSTM-based models. For 8-day runoff forecasts, our model achieves an NSE of 0.536, compared to 0.326 from the closest competitor. The integration of GAN-based feature extraction with spatial attention mechanisms improves predictive accuracy, particularly for peak-flow events. This method offers a powerful solution for addressing current challenges in water resource management and disaster planning under extreme climate conditions.

Semi‐implicit Lagrangian Voronoi approximation for the incompressible Navier–Stokes equations

AbstractWe introduce semi‐implicit Lagrangian Voronoi approximation (SILVA), a novel numerical method for the solution of the incompressible Euler and Navier–Stokes equations, which combines the efficiency of semi‐implicit time marching schemes with the robustness of time‐dependent Voronoi tessellations. In SILVA, the numerical solution is stored at particles, which move with the fluid velocity and also play the role of the generators of the computational mesh. The Voronoi mesh is rapidly regenerated at each time step, allowing large deformations with topology changes. As opposed to the reconnection‐based Arbitrary‐Lagrangian‐Eulerian schemes, we need no remapping stage. A semi‐implicit scheme is devised in the context of moving Voronoi meshes to project the velocity field onto a divergence‐free manifold. We validate SILVA by illustrative benchmarks, including viscous, inviscid, and multi‐phase flows. Compared to its closest competitor, the Incompressible Smoothed Particle Hydrodynamics method, SILVA offers a sparser stiffness matrix and facilitates the implementation of no‐slip and free‐slip boundary conditions.

Plant size inequality: How to analyse competing spatial dissimilarity indices?

Worldwide biodiversity loss is perceived as a major threat and is likely to be enforced by ongoing climate change. Continued monitoring of biodiversity can assist in compensating for decreasing biodiversity by goal-oriented conservation management. Plant size diversity, also termed size inequality and size hierarchy, has often been neglected in studies of biodiversity, but plays a crucial role in many ecosystems, e.g. in forests. Several competing diversity indices and other characteristics of spatial plant size inequality have been proposed but to date no protocols or guidelines exist for evaluating their relative merits. In our study, we proposed a broad framework for such an analysis. In order to put this framework to a test, we revisited the dissimilarity coefficient, a somewhat ignored but promising spatial size inequality index published at the end of the 1990s, identified its nearest index competitors and analysed their performances in different mathematical-statistical contexts. We learned that the dissimilarity coefficient is more sensitive in statistical significance tests relating to three very different summary characteristics than its closest competitor, the size differentiation index. In addition the dissimilarity coefficient has a more solid foundation than the differentiation index, since it is based on the well-known coefficient of variation. The dissimilarity coefficient can be recommended in a wide range of plant diversity applications. The principle of the dissimilarity coefficient can also be used to define an effective mark correlation function. We recommend our evaluation framework for use in similar analyses of competing diversity characteristics.

A multi-level action coupling reinforcement learning approach for online two-stage flexible assembly flow shop scheduling

Multi-product centralized delivery and kitting assembly present significant challenges to hierarchical co-processing in multi-stage manufacturing systems. The combinations of priority dispatching rules at each level are transiently adaptive, and the performance in online scheduling deteriorates rapidly with changing environment. This paper investigates the selection of rule combinations for sustained high-performance responsive scheduling in two-stage flexible assembly flow shop scheduling problem with asynchronous execution and complex decision correlation. A Multi-Level Action Coupling Deep Q-Network (MALC-DQN) approach is proposed for adaptive integrated scheduling in hybrid processing and assembly shops. Firstly, the problem is skillfully established as an event-triggered integrated decision markov decision process. The prioritized batch experience replay mechanism is employed to retain the complete correlation information of key decision sequences. Then, coupling and sequence feature extraction modules are developed to enhance the agent’s ability to perceive execution process and the environment. Furthermore, the multi-level wait-limit mechanism and efficient action filtering mechanism are designed to mitigate ineffective waiting waste and action space explosion during learning. Finally, a series of sophisticated experiments are conducted to validate the effectiveness of the proposed methodology. In 20 actual instances of different sizes, MLAC-DQN outperformed its closest competitor, with a 26.6% improvement in average tardiness. Moreover, extraordinary robustness is demonstrated in 16 sets of experiments involving different configurations of resources, orders, and arrival concentration levels.

ICA-CRMAS: Intelligent Context-Awareness Approach for Citation Recommendation based on Multi-Agent System

Navigating the ever-expanding sea of scientific literature presents a daunting challenge for researchers seeking relevant and up-to-date information. Traditional citation recommendation systems, while well-intentioned, often fall short due to their limited focus on text-based features and lack of contextual awareness. In this paper we introduce the ICA-CRMAS (Intelligent Context-Aware Approach for Citation Recommendation based on Multi-Agent System), an intelligent system that leverages the power of deep learning, semantic analysis, and multimodal learning to overcome these limitations. ICA-CRMAS goes beyond the surface, delving into the rich tapestry of information within academic papers, including figures, which often hold vital contextual clues. By weaving this contextual data directly into its recommendation models, ICA-CRMAS generates highly personalized and relevant suggestions. This comprehensive approach unlocks enhanced accuracy, diversity, and serendipity, enabling researchers to effectively discover papers aligning with their interests and research objectives. ICA-CRMAS illuminates its reasoning. Instead of opaque suggestions, the system provides clear explanations that justify and illustrate recommended citations. This transparency builds user confidence, allowing researchers to critically engage with and trust the system’s recommendations. Evaluation experiments conducted on real-world academic datasets demonstrate that ICA-CRMAS outperforms existing approaches across various metrics. it surpassing its closest competitor by a margin of 7.53 on accuracy, 6.07% on MRR and by 5.87 on Recall. User feedback further reinforces its effectiveness, with an Overall System Usability Scale (SUS) score of 76.73, exceeding benchmark scores for comparable systems.

On Representation Learning-based Methods for Effective, Efficient, and Scalable Code Retrieval

Code retrieval consists of finding relevant code snippets regarding a programmer’s query — an increasingly important task due to software ubiquity. Although significant progress has been made, there is still a need for refinement and improvement in current code retrieval solutions. Besides effective, we argue that solutions for the code retrieval task should be efficient and scalable to ever-increasing large code repositories. This paper introduces xCoFormer, the first representation learning-based model for the code retrieval task in the literature, which meets all these requirements. Our contributions include: (i) an interactive tag-training method that efficiently places a query closer to its relevant codes in the embedding space and (ii) use of a specialized loss function (N-pair) better suited to the code retrieval task. To evaluate our proposal, we conducted experiments with different versions of xCoFormer against several state-of-the-art deep learning models in datasets with different properties. In our experiments, xCoFormer produced the best cost-effectiveness tradeoff — it is as effective as our closest competitor while being thousands of times faster at search time. xCoFormer’s attention mechanism also (partially) meets the desirable requirement of explainability by exposing the “reasoning” of the model’s predictions. Finally, when used as a “proxy” method, fine-tuned with domain-specific code language models, such as Unixcoder and CodeBERT, our proposal achieves gains of more than 33% in effectiveness when compared to the use of general purpose language models such as BERT and RoBERTa.

PCA-ICA-LSTM: A Hybrid Deep Learning Model Based on Dimension Reduction Methods to Predict S&P 500 Index Price

AbstractIn this paper, we propose a new hybrid model based on a deep learning network to predict the prices of financial assets. The study addresses two key limitations in existing research: (1) the lack of standardized datasets, time scales, and evaluation metrics, and (2) the focus on prediction return. The proposed model employs a two-stage preprocessing approach utilizing Principal Component Analysis (PCA) for dimensionality reduction and de-noising, followed by Independent Component Analysis (ICA) for feature extraction. A Long Short-Term Memory (LSTM) network with five layers is fed with this preprocessed data to predict the price of the next day using a 5 day time horizon. To ensure comparability with existing literature, experiments employ an 18 year dataset of the Standard & Poor's 500 (S&P500) index and include over 40 technical indicators. Performance evaluation encompasses six metrics, highlighting the model's superiority in accuracy and return rates. Comparative analyses demonstrate the superiority of the proposed PCA-ICA-LSTM model over single-stage statistical methods and other deep learning architectures, achieving notable improvements in evaluation metrics. Evaluation against previous studies using similar datasets corroborates the model's superior performance. Moreover, extensions to the study include adjustments to dataset parameters to account for the COVID-19 pandemic, resulting in improved return rates surpassing traditional trading strategies. PCA-ICA-LSTM achieves a 220% higher return compared to the “hold and wait” strategy in the extended S&P500 dataset, along with a 260% higher return than its closest competitor in the comparison. Furthermore, it outperformed other models in additional case studies. Graphical Abstract

Multi-class boosting for the analysis of multiple incomplete views on microbiome data

BackgroundMicrobiome dysbiosis has recently been associated with different diseases and disorders. In this context, machine learning (ML) approaches can be useful either to identify new patterns or learn predictive models. However, data to be fed to ML methods can be subject to different sampling, sequencing and preprocessing techniques. Each different choice in the pipeline can lead to a different view (i.e., feature set) of the same individuals, that classical (single-view) ML approaches may fail to simultaneously consider. Moreover, some views may be incomplete, i.e., some individuals may be missing in some views, possibly due to the absence of some measurements or to the fact that some features are not available/applicable for all the individuals. Multi-view learning methods can represent a possible solution to consider multiple feature sets for the same individuals, but most existing multi-view learning methods are limited to binary classification tasks or cannot work with incomplete views.ResultsWe propose irBoost.SH, an extension of the multi-view boosting algorithm rBoost.SH, based on multi-armed bandits. irBoost.SH solves multi-class classification tasks and can analyze incomplete views. At each iteration, it identifies one winning view using adversarial multi-armed bandits and uses its predictions to update a shared instance weight distribution in a learning process based on boosting. In our experiments, performed on 5 multi-view microbiome datasets, the model learned by irBoost.SH always outperforms the best model learned from a single view, its closest competitor rBoost.SH, and the model learned by a multi-view approach based on feature concatenation, reaching an improvement of 11.8% of the F1-score in the prediction of the Autism Spectrum disorder and of 114% in the prediction of the Colorectal Cancer disease.ConclusionsThe proposed method irBoost.SH exhibited outstanding performances in our experiments, also compared to competitor approaches. The obtained results confirm that irBoost.SH can fruitfully be adopted for the analysis of microbiome data, due to its capability to simultaneously exploit multiple feature sets obtained through different sequencing and preprocessing pipelines.

Competition and Collaboration in Crowdsourcing Communities: What Happens When Peers Evaluate Each Other?

Crowdsourcing has evolved as an organizational approach to distributed problem solving and innovation. As contests are embedded in online communities and evaluation rights are assigned to the crowd, community members face a tension: They find themselves exposed to both competitive motives to win the contest prize and collaborative participation motives in the community. The competitive motive suggests they may evaluate rivals strategically according to their self-interest, the collaborative motive suggests they may evaluate their peers truthfully according to mutual interest. Using field data from Threadless on 38 million peer evaluations of more than 150,000 submissions across 75,000 individuals over 10 years and two natural experiments to rule out alternative explanations, we answer the question of how community members resolve this tension. We show that as their skill level increases, they become increasingly competitive and shift from using self-promotion to sabotaging their closest competitors. However, we also find signs of collaborative behavior when high-skilled members show leniency toward those community members who do not directly threaten their chance of winning. We explain how the individual-level use of strategic evaluations translates into important organizational-level outcomes by affecting the community structure through individuals’ long-term participation. Although low-skill targets of sabotage are less likely to participate in future contests, high-skill targets are more likely. This suggests a feedback loop between competitive evaluation behavior and future participation. These findings have important implications for the literature on crowdsourcing design, and the evolution and sustainability of crowdsourcing communities. Funding: This work was supported by the National Science Foundation [Grant IIS-1514283] and the U.S. Office of Naval Research [Grant N00014-17-1-2542]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/orsc.2021.15163 .

Mesocarnivores vary in their spatiotemporal avoidance strategies at communications hubs of an apex carnivore.

Mesocarnivores face interspecific competition and risk intraguild predation when sharing resources with apex carnivores. Within a landscape, carnivores across trophic levels may use the same communication hubs, which provide a mix of risks (injury/death) and rewards (gaining information) for subordinate species. We predicted that mesocarnivores would employ different strategies to avoid apex carnivores at shared communication hubs, depending on their trophic position. To test our prediction, we examined how different subordinate carnivore species in the Santa Cruz Mountains of California, USA, manage spatial overlap with pumas (Puma concolor), both at communication hubs and across a landscape-level camera trap array. We estimated species-specific occurrence, visitation rates, temporal overlap, and Avoidance-Attraction Ratios from camera traps and tested for differences between the two types of sites. We found that mesocarnivores generally avoided pumas at communication hubs, and this became more pronounced when pumas scent-marked during their most recent visit. Coyotes (Canis latrans), the pumas' closest subordinate competitor in our system, exhibited the strongest avoidance at communication hubs. Gray foxes (Urocyon cinereoargenteus) avoided pumas the least, which may suggest possible benefits from pumas suppressing coyotes. Overall, mesocarnivores exhibited various spatiotemporal avoidance strategies at communication hubs rather than outright avoidance, likely because they benefit from information gained while 'eavesdropping' on puma activity. Variability in avoidance strategies may be due to differential predation risks, as apex carnivores often interact more aggressively with their closest competitors. Combined, our results show how apex carnivores trigger complex species interactions across the entire carnivore guild and how trophic position determines behavioral responses and subsequent space use of subordinate mesocarnivores across the landscape.

A new over-dispersed count model based on Poisson-Geometric convolution

A new two-parameter discrete distribution, namely the PoiG distribution, is derived by the convolution of a Poisson variate and an independently distributed geometric random variable. This distribution generalizes both the Poisson and geometric distributions and can be used for modeling over-dispersed as well as equi-dispersed count data. A number of important statistical properties of the proposed count model, such as the probability generating function, the moment generating function, the moments, the survival function and the hazard rate function have been studied in detail. Monotonic properties such as the log concavity are also studied. Stochastic ordering for the proposed model is also investigated in detail. Method of moments and the maximum-likelihood estimators of the parameters of the proposed model are found to work well in simulation experiments as well as in the real-life data analysis. Asymptotic confidence intervals and the coverage probabilities of the parameters are also studied through extensive simulation experiments. Two real life modelling applications demonstrate that the proposed distribution may prove to be useful for the practitioners for modelling over-dispersed count data compared to its closest competitors.

Pemetaan Persepsi Konsumen untuk Meningkatkan Daya Saing Produk: Studi Kasus T-shirt Libre Everyday

In competing, it is important to understand consumer perceptions of products. Even though consumers are satisfied with the company's brand image, the company can still lose its customers. Companies need to identify consumer characteristics in their marketing strategy, because this can help formulate the right position for the product. Understanding the concept of product or brand positioning is very important to maintain the company's continuity in competition. This research aims to determine product positioning based on consumer perceptions of product attributes. Consumer perception mapping was carried out on the Libre Everyday t-shirt product using a purposive sampling method involving 70 respondents who had been t-shirt users for more than two years and were familiar with at least three specified t-shirt brands. The data obtained was processed using IBM SPSS 23 software to produce perceptual mapping. The calculation results show that this perception map model is feasible and acceptable, as proven by the goodness of fit test with an R-square value of 0.15312 and a stress value of 0.97814. This perception map depicts brand positioning, product attributes, and closest competitors. To increase product competitiveness, it is necessary to improve product positioning by utilizing superior product attributes, improving marketing communications, and adding marketing channels.

DnaDot - fixing ecology and evolution’s blind spot, population size

To understand and manage any ecosystem, one of the most critical indicators is census population size (‘Nc’) for each component species. This is why the International Union for Conservation of Nature (IUCN) stipulates that it is ideal to know census size to an accuracy of ±10 % for each managed, endangered or exploited population. Knowledge of population size is also critical for: ecological studies in general; allowing biodiversity measures with low error margins; and assessing potential for evolutionary adaptation in long-term management. Unfortunately, despite their importance, all existing methods for estimating Nc have many difficulties, especially requiring independent knowledge or assumptions about demography, including: immigration, emigration, family size and its variance. As a result, even in the IUCN red list, many species are listed despite having unknown population size or trends. Therefore, we need an improved approach. This article introduces ‘DnaDot’, a strong new addition to our armory of census population size estimates, because it is an accurate estimate, with few assumptions. DnaDot is based on mark-release-recapture, but instead of marking, uses pre-existing polymorphisms to divide the population into separate groups. The method uses one sample, minimizing effort, uses no demographic assumptions or data, and does not require genotyping to be accurate enough to identify individuals and kin, which is problematic for other genetic methods especially when degraded field samples are used. DnaDot applies to any kind of variants, from tissue samples, or from noninvasive samples such as hair provided that the researcher can identify species-identity and variants within the species. DnaDot outperformed close competitors on minimal assumptions, and good detectability of marks. Also, in simulations of a wide range of scenarios, DnaDot had superior accuracy and precision, usually meeting the IUCN criterion of ±10 %, which competing methods rarely achieved. Finally, one other method is said to need smaller sample sizes than DnaDot, but only if that method is not required to meet the performance criteria of DnaDot.

PathMap: a path-based mapping tool for long noisy reads with high sensitivity.

With the rapid development of single-molecule sequencing (SMS) technologies, the output read length is continuously increasing. Mapping such reads onto a reference genome is one of the most fundamental tasks in sequence analysis. Mapping sensitivity is becoming a major concern since high sensitivity can detect more aligned regions on the reference and obtain more aligned bases, which are useful for downstream analysis. In this study, we present pathMap, a novel k-mer graph-based mapper that is specifically designed for mapping SMS reads with high sensitivity. By viewing the alignment chain as a path containing as many anchors as possible in the matched k-mer graph, pathMap treats chaining as a path selection problem in the directed graph. pathMap iteratively searches the longest path in the remaining nodes; more candidate chains with high quality can be effectively detected and aligned. Compared to other state-of-the-art mapping methods such as minimap2 and Winnowmap2, experiment results on simulated and real-life datasets demonstrate that pathMap obtains the number of mapped chains at least 11.50% more than its closest competitor and increases the mapping sensitivity by 17.28% and 13.84% of bases over the next-best mapper for Pacific Biosciences and Oxford Nanopore sequencing data, respectively. In addition, pathMap is more robust to sequence errors and more sensitive to species- and strain-specific identification of pathogens using MinION reads.

The theory of argument formation: between kinds and properties

Abstract Chierchia (1998) developed a cross-linguistic extension to Carlsons seminal work on bare nouns (BNs), producing the most influential theory of argument formation to date, henceforth the Kinds Approach (KA). The core achievements of the KA included the derivation of the generalized narrow scope behavior of BNs and of the existence of generalized classifier languages. There are cracks in the picture, though. The narrow scope behavior of BNs is more fine-grained than is generally assumed and the KA lacks the flexibility to deal with it (Le Bruyn & Swart 2022). The appeal of the KAs derivation of the existence of generalized classifier languages heavily relied on all nouns in these languages being mass-like, an assumption that has since been abandoned (Chierchia 2010; Jiang 2020). These developments call for a reassessment of the KA and one of its closest competitors: Krifka 2003. Krifka assumes nouns never start life as kinds but as predicates, leading us to qualify his approach as a Properties Approach (PA).We adopt a translation corpus approach and assess the explanatory potential of the KA and the PA by comparing the distribution of BNs and related expressions in (in)definite contexts across six typologically different languages. Our results show that the PA has a distinct advantage over the KA and identify pseudo-incorporation and the way it varies across languages as a primary focus for future research.

Analysis of Thailand's fixed broadband internet services demand: A merger screening of AIS and 3BB

The article conducted a ‘merger screening’ of the potential impact of the merger between AIS and 3BB, which were the second and third largest internet service providers among the four major operators prior to the merger in 2022. National survey data across country in 2022 and discrete choice model was used for the analysis. Surprisingly, finding showed that AIS and 3BB were not the closest competitors, but instead True, the provider with the largest market share, was their closest competitor. Based on the pre-merger analysis, it was predicted that service prices were likely to increase, and this merger is comprised a relevant market. To accurately define the market (market definition), other operators' services should not be considered as well. Considering the rising magnitude in service prices, it is estimated that, on average, prices will increase by 7.51%–31.51% compared to the current rates in 2022. Moreover, the study revealed that the merger would lead to an average increase in cost efficiency of 15.01% of the current marginal cost, which would allow the post-merger service prices to remain unchanged if the cost savings are passed on to consumers. A key policy recommendation was for regulatory agencies to conduct further analysis of competition, including evaluating entry barriers, the response of non-merging competitors, and competition without regulated pricing. Additionally, the study examines the potential future impacts on social welfare.

АНАЛИЗ ОТРАСЛЕВОЙ АНИЗОТРОПИИ В МЕЖДУНАРОДНОЙ ЛЕГКОЙ ПРОМЫШЛЕННОСТИ

In the interests of strategic positioning of enterprises and in order to ensure conditions for their effective competition, it is necessary to clarify the enterprise’s place on the industry map in the “resilience – efficiency” coordinates. Such clarification is possible when studying a new property of sectoral anisotropy. Using the example of international light industry, we show that all enterprises in the industry can be divided into three large classes: “pioneers”, “miners” and “convoys”, with each class dominating each other in the Pareto sense. The current positioning of the enterprise involves clarifying the layer in which it is located, as well as the list of its closest competitors, highlighting the strengths and weaknesses of the company under study. During anisotropy analysis, methods of fuzzy set theory and soft computing are used, including obtaining an industry stability zone in the form of an R-lens.

TriCI: Triple Cross-Intra Branch Contrastive Learning for Point Cloud Analysis.

Whereas contrastive learning eliminates the need for labeled data, existing methods may suffer from inadequate features due to the conventional single shared encoder structure and struggle to fully harness the rich spectrum of 3D augmentations. In this paper, we propose TriCI, a self-supervised method that designs a triple-branch contrastive learning architecture. During contrastive pre-training, we generate three augmented versions of each input point cloud sample and pair each augmented sample with the original one, resulting in three unique positive pairs. We subsequently feed the pairs into three distinct encoders, each of which extracts features from its corresponding input positive pair. We design a novel cross-branch contrastive loss and use it along with the intra-branch contrastive loss to jointly train our network. The proposed cross-branch loss effectively aligns the output features from different perspectives for pre-training and facilitates their integration for downstream tasks, particularly in object-level scenarios. The intra-branch loss helps maximize the feature correspondences within positive pairs. Extensive experiments demonstrate the superiority of our TriCI in self-supervised learning, and show its strong ability in enhancing the performance of downstream object classification and part segmentation tasks. Interestingly, our TriCI achieves a 92.9% accuracy for linear SVM evaluation on ModelNet40, exceeding its closest competitor by 1.7% and even exceeding some supervised methods.