ラプラシアンカーネルによるハブの解消 (original) (raw)
ラプラシアンカーネルによるハブの解消
2013, Transactions of the Japanese Society for Artificial Intelligence
A "hub" is an object closely surrounded by, or very similar to, many other objects in the dataset. Recent studies by Radovanović et al. demonstrated that in high dimensional spaces, objects close to the data centroid tend to become hubs. In this paper, we show that the family of kernels based on the graph Laplacian makes all objects in the dataset equally similar to the centroid, and thus they are expected to make less hubs when used as a similarity measure. We investigate this hypothesis using both synthetic and real-world data. It turns out that these kernels suppress hubs in some cases but not always, and the results seem to be affected by the size of the data-a factor not discussed previously. However, for the datasets in which hubs are indeed reduced by the Laplacian-based kernels, these kernels work well in classification and information retrieval tasks. This result suggests that the amount of hubs, which can be readily computed in an unsupervised fashion, can be a yardstick of whether Laplacian-based kernels work effectively for a given data.
Related papers
Transactions of the Japanese Society for Artificial Intelligence, 2016
The Generalized Mutual Assignment Problem (GMAP) is a maximization problem in distributed environments, where multiple agents select goods under resource constraints. Distributed Lagrangian Relaxation Protocols (DisLRP) are peer-to-peer communication protocols for solving GMAP instances. In DisLRPs, agents seek a good quality upper bound for the optimal value by solving the Lagrangian dual problem, which is a convex minimization problem. Existing DisLRPs exploit a subgradient method to explore a better upper bound by updating the Lagrange multipliers (prices) of goods. While the computational complexity of the subgradient method is very low, it cannot detect the fact that an upper bound converges to the minimum. Moreover, solution oscillation sometimes occurs, which is one of the critical issues for the subgradient method. In this paper, we present a new DisLRP with a Bundle Method and refer to it as Bundle DisLRP (BDisLRP). The bundle method, which is also called the stabilized cutting planes method, has recently attracted much attention as a way to solve Lagrangian dual problems in centralized environments. We show that this method can also work in distributed environments. We experimentally compared BDisLRP with Adaptive DisLRP (ADisLRP), which is a previous protocol that exploits the subgradient method, to demonstrate that BDisLRP performed convergence faster with better quality upper bounds than ADisLRP.
HFIR 照射された低放射化フェライト鋼の照射下クリープ挙動
Journal of the Japan Institute of Metals, 2007
The irradiation creep behavior of F82H and several JLF 1 steels has been measured at 573 and 773 K up to 5 dpa using pressurized creep tubes irradiated in HFIR. These tubes were pressurized with high purity helium to hoop stress levels of 0 to 400 MPa at the irradiation temperature. The results for F82H and JLF 1 with hoop stress of 400 MPa showed small creep strains (＜0.25) after irradiation at 573 K. The irradiation creep strain at 573 K in these steels is linearly dependent on the applied stress at stress levels below 250 MPa. However, at higher hoop stress levels, the creep strain is nonlinear. At 773 K, the irradiation creep strain of F82H is linearly dependent on the applied hoop stress level below 100 MPa. At higher stress levels, the creep strain increased strongly. The creep compliance co efficient for F82H and JLF 1 is consistent with the values obtained for other steels. These data contribute to the materials database for ITER test blanket design work.
Japanese Journal of Neurosurgery, 2014
The defining characteristics of patients for whom craniotomy is the recommended treatment for their refractory seizures have been clarified during the last two decades. Seizure outcome of resection surgery for focal seizures associated with focal lesions is satisfactory. Particularly for mesial temporal lobe epilepsy, surgical treatment should be considered from the earlier stage of the disease. Meanwhile, surgical outcomes in extratemporal lobe epilepsy and nonlesional epilepsy still have room for improvement using various approaches. Although not curative, palliative procedures and neuromodulation therapies are available for these truly intractable epilepsy cases at present.
ラット上顎臼歯におけるBMP-石灰藻由来ハイドロキシアパタイト複合体の修復象牙質誘導
2010
rhBMP(骨形成タンパク質)v-FRIOS Algipore(石灰藻由来の多孔質ハイドロキシアパタイト)複合体がラットを用いた直接覆髄実験において効果的に良質な修復象牙質形成を誘導するか検討した。BMP-HA群は、HA群、水酸化カルシウム群に比べて早期から旺盛な修復象牙質形成を認め、細管構造を有する緻密な構造であった。BMP-HA群では早期から炎症が軽度で、全実験期間を通して他の実験群に比べて歯髄の炎症の程度は軽度であった。水酸化カルシウム製剤によって誘導された修復象牙質は全体的に多孔質で、トンネル状の欠陥が存在し、全ての観察期間を通して歯髄に炎症反応を認めた。歯髄保存療法剤としてのrhBMPv-FRIOS Algipore複合体の有効性が示唆された
DENKI-SEIKO[ELECTRIC FURNACE STEEL], 2001
In October 1998, a new mass-produced car with titanium engine-valves was released from TOYOTA Motor Corporation. Both intake and exhaust valves were manufactured via a newly developed cost-effective P/M forging process. Furthermore, the material which was specially designed for the exhaust one is a unique TiB reinforced titanium metal matrix composite (MMC). This paper discusses the materials and manufacturing methods used. The tensile, fatigue strength and creep resistance of the developed MMC are always superior to those of the typical heat-resistant steel of 21-4N.
Loading Preview
Sorry, preview is currently unavailable. You can download the paper by clicking the button above.