An Efficient Hybrid Music Recommender System Using an Incrementally Trainable Probabilistic Generative Model

This paper presents a hybrid music recommender system that ranks musical pieces while efficiently maintaining collaborative and content-based data, i.e., rating scores given by users and acoustic features of audio signals. This hybrid approach overcomes the conventional tradeoff between recommendation accuracy and variety of recommended artists. Collaborative filtering, which is used on e-commerce sites, cannot recommend nonbrated pieces and provides a narrow variety of artists. Content-based filtering does not have satisfactory accuracy because it is based on the heuristics that the user's favorite pieces will have similar musical content despite there being exceptions. To attain a higher recommendation accuracy along with a wider variety of artists, we use a probabilistic generative model that unifies the collaborative and content-based data in a principled way. This model can explain the generative mechanism of the observed data in the probability theory. The probability distribution over users, pieces, and features is decomposed into three conditionally independent ones by introducing latent variables. This decomposition enables us to efficiently and incrementally adapt the model for increasing numbers of users and rating scores. We evaluated our system by using audio signals of commercial CDs and their corresponding rating scores obtained from an e-commerce site. The results revealed that our system accurately recommended pieces including nonrated ones from a wide variety of artists and maintained a high degree of accuracy even when new users and rating scores were added.     

Difference in the detection limits of flaws in the depths of multi-layered and continuous aluminum plates using low-frequency eddy current testing

We examined the difference in the detection limits of flaws in the depths of multi-layered and continuous aluminum plates using low-frequency eddy current testing. The detection limits were measured by using a magneto-resistive sensor. Comparing the frequency of an applied magnetic field, the detection limit at 50 Hz is deeper than that at 1 kHz. Comparing the sample structure, the detection limit in the multi-layered samples is deeper than that in the continuous samples. These results are likely due to the differences in the skin depth and conductivity of the sample.     

Sinusoidal endothelial cell damage by activated macrophages in rat liver necrosis

BACKGROUND: Massive hepatic necrosis caused by fibrin deposition in the hepatic sinusoids develops with hepatic macrophage activation in rats given endotoxin after administration of heat-killed Corynebacterium parvum. Targeted cells of such macrophages were investigated. METHODS: In C. parvum-treated rats, the pathological appearance of liver cells was serially measured in serum following endotoxin administration and compared with the appearance in the perfusate during closed liver perfusion with endotoxin. RESULTS: Serum activities of tumor necrosis factor, purine nucleoside phosphorylase present in both hepatocytes and sinusoidal endothelial cells, and levels of alanine aminotransferase were higher after 30 minutes, 1 hour, and 3 hours, respectively. Pretreatment of rats with gadolinium chloride, an inhibitor of macrophage function, reduced this liver injury. Although alanine aminotransferase activity remained almost unchanged in the liver perfusate, purine nucleoside phosphorylase activity increased. This increase was reduced when rats were pretreated with gadolinium chloride. There was sinusoidal endothelial cell damage around hepatic macrophages in the liver perfused with endotoxin. CONCLUSIONS: Activated hepatic macrophages may cause sinusoidal endothelial cell damage leading to hepatocyte necrosis in rats given C. parvum and endotoxin.     

Regulation of CD4 T cell reactivity to self and non-self

While the thymus may be effective in inducing tolerance to lymphoid associated antigens, it is not as efficient in deleting T cells reactive to peripheral tissue specific antigens. Therefore, to maintain self tolerance to peripheral tissues, post-thymic mechanisms must be invoked. One important way to prevent autoimmune pathology mediated by autoreactive CD4 T cells is the diversion of clones to regulatory Th2 effector cells. However, many different factors contribute in vivo to the decision of stimulated CD4 T cells to develop into Th1 versus Th2 cells. For example, T cell signaling pathways may influence the types of cytokines produced by naive T cells, and studies have provided evidence for a genetic polymorphism among common mouse strains that can significantly influence the early cytokine production in stimulated naive CD4 T cells. The allele carried by the BALB/c strain promotes IL-4 production, and consequently provides resistance to autoimmune diabetes in our transgenic mouse model. In addition, antigen presenting cells can influence the development of stimulated CD4 T cells in part through the production of cytokines such as IL-12. The absorption of IL-12 in vivo can permit the expansion of Th2 type effector cells, and this phenomenon will also protect mice from autoimmunity. Finally, the relative potency of various class II positive antigen presenting cell types can influence the development of autoreactive T cells, with dendritic cells apparently being the strongest stimulator of Th1 responses. Consistent with this notion, a relB knockout mouse, which is missing dendritic cells, appears to drive Th2 development even in response to viral infection. In sum, these various influences over the Th1/Th2 decision in vivo may provide new targets for immunotherapy of autoimmune diseases.     

Tunnel effect in GaInAsP/InP double heterostructure diode

Negative differential resistance in a GaInAsP/InP double heterostructure diode with a pn junction annealed at 550-600°C is demonstrated. It is found that negative differential resistance is caused by the diffusion of impurities, Zn and Te, into the undoped GaInAsP layer     

Selective reduction of nitrogen oxides with various organic substances on precious metal catalysts under a high GHSV condition

Performances of Pt and Rh catalysts for the selective reduction of NO were investigated using various reducing agents under high gas hourly space velocity (GHSV). Higher activities were attained when olefins were used for both Pt and Rh catalysts, and when oxygen containing substances such as methyl tert.-butyl ether (MTBE) were used in the case of Rh catalysts. Selectivity toward N₂ for the NO converted was from 7% to 38% for Pt catalysts and from 45% to 67% for Rh catalysts. A comparison of N₂O decomposition rates on these catalysts has shown that gaseous N₂O is not a principal intermediate in the pathway for N₂ formation. The high selectivity toward N₂ for Rh/Al₂O₃ at low GHSV conditions is attributed to an additional catalytic effect of A1₂O₃, which can reduce NO₂ with propylene at temperatures as low as 350°C.     

Ionic conductivity of polymer electrolytes and future applications

Polymer electrolytes are solvent-free ion-conducting polymers and provide new and attractive materials in both polymer chemistry and electrochemistry. A proper understanding of ion dissociation and ion transport in such polymers is necessary for their application as solid electrolytes in electrochemical devices. Ionic conduction behaviour in polymer electrolytes is described here in relation to the characteristic properties. Of special interest is the ability of polymer electrolytes to include various kinds of electroactive molecules within them. The combination of this ability with their high ionic conductivity has enabled polymer electrolytes to be used as media for electrochemical syntheses and redox reactions.     

Elevated interleukin-8 levels in the urine of children with renal scarring and/or vesicoureteral reflux

PURPOSE: Elevation of urinary levels of interleukin-6 and 8 has been observed in patients with acute urinary tract infections. However, to our knowledge there have been no studies concerning the secretion of interleukin-6 and 8 into the urine after acute inflammation has resolved and renal scarring has occurred. On the other hand, it is well known that cytokines are variously related to glomerular diseases and, thus, it is possible that the progression of reflux nephropathy depends on interleukin-6 or 8. Therefore, we assessed urinary levels of interleukin-6 and 8 in children with vesicoureteral reflux and/or renal scarring. MATERIALS AND METHODS: We evaluated interleukin-6 and interleukin-8 levels in the urine of 32 children without a urinary tract infection who presented or were admitted to our hospital because of vesicoureteral reflux between April and December 1994. Interleukin-6 and 8 were determined using a commercially available human enzyme-linked immunosorbent assay kit and the 2-step sandwich method. RESULTS: Urinary interleukin-6 levels were below the lower detection limit (less than 10 pg./ml.) in all samples. There were statistically significant differences between urinary interleukin-8 levels in children with and without renal scarring (p = 0.001), and with and without vesicoureteral reflux (p = 0.0246). CONCLUSIONS: Urinary interleukin-8 is an effective marker for renal scarring and vesicoureteral reflux.