Search

The training data distribution is often biased towards objects in certain orientations and illumination conditions. While humans have a remarkable capability of recognizing objects in out-of-distribution (OoD) orientations and illuminations, Deep Neural Networks (DNNs) severely suffer in this case, even when large amounts of training examples are available. Neurons that are invariant to orientations and illuminations have been proposed as a neural mechanism that could facilitate OoD generalization, but it is unclear how to encourage the emergence of such invariant neurons. In this paper, we investigate three different approaches that lead to the emergence of invariant neurons and substantially improve DNNs in recognizing objects in OoD orientations and illuminations. Namely, these approaches are (i) training much longer after convergence of the in-distribution (InD) validation accuracy, i.e., late-stopping, (ii) tuning the momentum parameter of the batch normalization layers, and (iii) enforcing invariance of the neural activity in an intermediate layer to orientation and illumination conditions. Each of these approaches substantially improves the DNN's OoD accuracy (more than 20% in some cases). We report results in four datasets: two datasets are modified from the MNIST and iLab datasets, and the other two are novel (one of 3D rendered cars and another of objects taken from various controlled orientations and illumination conditions). These datasets allow to study the effects of different amounts of bias and are challenging as DNNs perform poorly in OoD conditions. Finally, we demonstrate that even though the three approaches focus on different aspects of DNNs, they all tend to lead to the same underlying neural mechanism to enable OoD accuracy gains - individual neurons in the intermediate layers become invariant to OoD orientations and illuminations. We anticipate this study to be a basis for further improvement of deep neural networks' OoD generalization performance, which is highly demanded to achieve safe and fair AI applications.

All-solid-state Li batteries are suitable for the application of 5 V-class cathode materials that can provide high energy densities. However, it is hard to fabricate bulk-type batteries composed of a mixture of active materials and solid electrolytes because of the difficulty of ensuring a conductive path and suppressing the formation of reactive layers at the interface. Herein, we report a simple mechanical processing route to composite cathode granules synthesized from a 5 V-class LiCoPO4 cathode, conductive carbon, and phosphate solid electrolyte. The homogeneously dispersed component particles were granulated during the mechanical treatment. The first discharge capacity is 82 mAh/g for the bulk-type all-solid-state battery obtained by using a composite cathode pellet, polymer solid electrolyte, and Li anode. Scanning transmission electron microscope-electron energy loss spectroscopy measurements did not reveal any Co diffusion or the formation of a resistive layer at the interface between LiCoPO4 and the solid electrolyte. The fabrication of a composite cathode by mechanical treatment is essential for the development of high-energy bulk-type all-solid-state batteries.

ObjectivesOur 2007 study of 32 patients with ovarian cancer reported the possible involvement of tissue factor (TF) in the development of venous thromboembolism (VTE) before treatment, especially in clear cell carcinoma (CCC). This follow-up study further investigated this possibility in a larger cohort.MethodsWe investigated the intensity of TF expression (ITFE) and other variables for associations with VTE using univariate and multivariate analyses in 128 patients with epithelial ovarian cancer initially treated between November 2004 and December 2010, none of whom had received neoadjuvant chemotherapy. Before starting treatment, all patients were ultrasonographically screened for VTE. The ITFE was graded based on immunostaining of surgical specimens.ResultsHistological types were serous carcinoma (n = 42), CCC (n = 12), endometrioid carcinoma (n = 15), mucinous carcinoma (n = 53), and undifferentiated carcinoma (n = 6). The prevalence of VTE was significantly higher in CCC (34%) than in non-CCC (17%, P = 0.03). As ITFE increased, the frequencies of CCC and VTE increased significantly (P < 0.001 and P = 0.014, respectively). Multivariate analysis identified TF expression and pretreatment dimerized plasmin fragment D level as significant independent risk factors for VTE development. These factors showed particularly strong impacts on advanced-stage disease (P = 0.021).ConclusionsThe 2007 cohort was small, preventing multivariate analysis. This study of a larger cohort yielded stronger evidence that the development of VTE in epithelial ovarian cancer may involve TF expression in cancer tissues.

Lithium Ion Conducting Glass Ceramic (LICGCTM) with a NASICON type structure was developed using industrial glass melting methods and functions as a solid lithium ion electrolyte.1-3) The ionic conductivity of the glass ceramic with Li1+x+yAlxTi2–xSiyP3–yO12 specific composition has values equal to or greater than 1×10-3 Scm-1 at room temperature. The NASICON type structure of LICGCTM was formed and crystalized in the starting bulk amorphous glass during heat treatment. The resulting glass ceramic has high chemical resistance and does not lose weight in boiling distilled water. The ionic conductivity of this electrolyte does not deteriorate, even if soaked in an aqueous of Li salt solution4-5. These characteristics indicate LICGCTM might be used to partially replace organic electrolytes and to enhance the properties of lithium ion secondary batteries. The effects of using LICGCTM in powder form as a cathode additive for lithium ion secondary batteries were investigated by OHARA is presented in this paper. LICGCTM powder of a controlled particle size, 0.4μm(D50), was prepared to fit into the inter-particle gaps of active materials generally used in a cathode. The cathode slurry was prepared by mixing standard cathode component materials (active material, carbon, binder) with LICGCTM powder and NMP. The slurry was then coated onto Al foil to form the cathode. The electrochemical performance of the NMC cathode, when evaluated in a half cell, indicated a capacity increase of 21% at 3C discharge and a reduced charging time of 20% at CC-CV condition, compared to a standard cathode with equal capacity. A full cell was fabricated using the cathode containing the LICGCTM powder additive and a carbon anode. The DC resistance of the full cell at DOD50 was decreased by more than 20%. We confirmed that LICGCTM powder when used as a cathode additive is effective in enhancing the performance of lithium ion secondary batteries. OHARA proceeded to evaluate battery retention capacity at 60℃. At this temperature, the charge and discharge cycle numbers of a lithium ion battery containing 1% LICGCTM powder in the NMC cathode increased by 4 times, at the point where the retention capacity is typically decreased by 10%, and the battery capacity at 300 cycles was increased by 14%. In the case of a lithium ion battery containing 1% LICGCTM powder additive in the LCO cathode at 60℃, the retention capacity was increased by 13% at 300 cycles. It was determined the 1% LICGCTM powder additive in the cathode inhibits the generation of the resistive layer and X-ray absorption spectroscopy (XAS, Ritsumeikan Univ. SR center) indicated the improved surface state of the LCO active material resulted in longer battery lifetime. OHARA’s results indicate that cathodes containing LICGCTM glass ceramic electrolyte powder can significantly improve battery retention capacity, by controlling the degradation of the cathode, when charging and exposed at high temperature. References [1] J. Fu, Solid State Ionics 96 (1997) 195-200 [2] J. Fu, J. Material Science 33 (1998) 1549-1553 [3] Y. Inda, T. Katoh, and M. Baba, J. Power Sources 174 (2007) 741 [4] N. Imanishi, S. Hasegawa, T. Zhang, A. Hirano, Y. Takeda and O. Yamamoto, J. Power Sources 185 (2008)1392 [5] S. Hasegawa, N. Imanishi, T. Zhang, J. Xie, A. Hirano, Y. Takeda and O. Yamamoto, J. Power Sources 189 (2009)371

Submicron particles of Na SuperIonic CONductor (NASICON)-type Li2O-Al2O3-TiO2-P2O5 (LATP) are possible to be densified at low temperature of 300 ºC by applying high pressure of 600 MPa during spark-plasma sintering (SPS) process. This process provides a novel route for assembling bulk-type all-solid-state lithium-ion battery (ASS-LIB), which promises high safety and high durability of energy storages since the origins of flammability and undesirable electrochemical events are prevented by using oxide as an electrolyte instead of the non-aqueous liquid. The main challenge for assembling bulk-type ASS-LIB are how to reduce the sintering temperature of the oxide electrolyte since most of the electrode materials decompose readily at high temperatures. Previously, we have attempted to design LiCoO2-electrode/NASICON-type LiTi2(PO4)3-electrolyte interface by conventional SPS process since quick/low-temperature treatment by the SPS has a possibility for reducing the decomposition kinetically.1 However, the several-μm ion-blocking interphase was easily-produced at the interface. Therefore, further low-temperature technique is required for the design of impurity-free electrode/electrolyte heterointerfaces. Commonly, the low-temperature sinterability of ceramic (or metal) particles could be accomplished with the increase of the contact for active surfaces by reducing particle size and/or applying higher pressure under sintering. Kinemuchi et al. has suggested that the low-temperature densification (LTD) of the oxide was performed by combining the use of nano particles and external pressure of over several-hundred MPa during the SPS.2 The LTDs for lithium-ion conductive oxides have possibilities for producing impurity-free electrode/electrolyte interface, and therefore would be one of the promising techniques for assembling oxide-based bulk-type ASS-LIB. In the present study, we used submicron size of NASICON-type Li2O-Al2O3-TiO2-P2O5 (LATP) powder as high bulk lithium-ion conductive oxides, which have been developed as Lithium Ion Conductive Glass Ceramics; LICGCTM, by OHARA INC..3,4 The densification of the LATP powder compact could be certified at low temperature of 200 - 250 °C under applying a pressure of 600 MPa during the SPS. Generally, the LTDs for oxides were achieved using not submicron particles but nano particles. However, the LTDs of the submicron-size particles could be confirmed in a phosphate group, which includes Li3PO4, AlPO4, and also LATP. Therefore, the bulk-type ASS-LIBs with phosphate-based electrodes such as LiFePO4 and Li2CoPO4F were successfully assembled by the LTD processes because the undesirable-thermal impurities at electrode/electrolyte interface were prevented during the low-temperature treatment. The characterization of LiFePO4/LATP interface and the electrochemical performance of ASS-LIB cell (Li/dry polymer/LATP/LiFePO4) will be presented in the conference. References [1] Y. Kobayashi et al., J. Power Sources, 1999, 81-82, 853. [2] Y. Kinemuchi et al., RSC Adv., 2016, 6, 24661. [3] J. Fu, Solid State Ionics, 1997,96, 195. [4] J. Fu, J. Mater. Sci., 1998, 33, 1549.

Lithium ion batteries are widely used in electric devices and electric vehicles due to their excellent performance at high power and the high energy density they provide compared to other conventional batteries. The application for electric vehicles is expanding worldwide as our society moves towards a lower carbon footprint but the lithium ion battery technology still needs further improvements in terms of physical properties, output, durability and safety. Current lithium ion batteries also degrade with charge–discharge cycles and storage time. In the cathode, low retention capacity, accompanied by the growth of resistive thin films and crystal structure deterioration on the surface of the active materials, remain unsolved problems. Modifications to the cathode surfaces have been researched and proven as effective measures to improve durability. Surface coatings on lithium‐ion battery cathodes improve their durability. Solid oxide materials like MgO, ZrO2 and SiO2 are applied or processed on to the cathode active materials by sol-gel, PVD or mechanochemical treatments and the solid oxide layer plays a role as an artificial solid electrolyte interface. The solid oxide coating layer prevents contact between the electrolyte and the active materials thus inhibiting the redox process and leading to an improvement in the cycle performance of the battery. We confirmed the LCO cathode containing a small % of Ohara lithium ion glass ceramic (LICGCTM ) PW-01 powder showed less deterioration and transition metal reduction at high temperature (60℃). The LICGCTM PW-01 powder was produced using a glass melting and ceraming method. A Li 2O-Al 2O 3-TiO 2-P 2O 5 glass was prepared and converted into glass-ceramics through heat-treatment. The lithium ion conductivity of the glass-ceramic with Li 1 + x Al x Ti 2 − x (PO 4) 3 phase is extremely high 1 × 10 −3 Scm −1 at room temperature. We found that adding LICGCTM PW-01 powder in the cathode had other significant effects, including increasing the battery capacity at high rate and at low temperature (-20℃). We also verified via by EIS measurement that the LICGCTM PW-01 powder additive significantly reduced the charge transfer resistance. The use of the LICGCTM PW-01 powder as a cathode slurry additive can serve an important role in improving the performance of lithium ion batteries.

The effects of the removal of the oily surface film on initial swim bladder inflation and survival in cultured seven-band grouper Hyporthodus septemfasciatus were investigated during the larval stage. Additionally, the relationship between swim bladder inflation failure and malformations in later stages was examined. To explore promotion conditions for inflation, duplicate experiments were conducted under three different water surface conditions: removal of the oily surface film (ROF group), covering the water surface with an oil film (COF group), and sealing the water surface with liquid paraffin (SLP group). Larval swim bladder inflation was first observed 14 days after hatching (DAH) in the ROF group; thereafter, swim bladder inflation rates increased to 11.1% and 38.7% until the end of the experiments. This indicates that initial swim bladder inflation occurs at the pre-flexion stage in larval development. Conversely, bladder inflation was rarely observed in the COF or SLP groups, indicating the promotional effect of surface film removal on larval swim bladder inflation and suggesting that gulping atmospheric air is necessary for initial inflation. Although larval water surface death occurred frequently in the ROF group, no significant difference existed in any group between the survival rates at 10 DAH and at the end of the experiment. The incidence rate of spinal curvature and deformation of the centra was significantly higher in individuals without initial swim bladder inflation, and these malformations mainly occurred in the caudal vertebra. Therefore, the removal of the oily surface film promotes initial swim bladder inflation and reduces larval malformations.

The present study focused on the influence of the apparent viscosity of activated sludge on mixed liquor oxygen transfer rates, and examined whether both could be modified by the addition of an organic polyelectrolyte. Batch experiments carried out using mixed liquor samples in a stirred tank reactor revealed that α -factors, which incorporate any difference in oxygen mass transfer coefficient values between clean water and mixed liquor samples, showed exponential relationships when plotted against increases in mixed liquor apparent viscosities and sludge mixed liquid suspended solids values. Furthermore, oxygen mass transfer coefficients increased exponentially with increases in the Reynolds number, a parameter which reflects liquid turbulence levels. Oxygen mass transfer rates of five sludge samples collected from four full scale wastewater treatment plants (WWTPs) also increased by about 10% after flocculation, while with a sludge sample collected from the lab scale reactor, it decreased by about 5%. One likely explanation is that the apparent viscosities of the WWTPs and the lab-reactor sludges decreased and increased respectively after flocculation. Such changes probably reflect changes in the volume fraction of mixed liquor suspended solids as indicated by changes in sludge volume indices, particle size values and endogenous respiration rates before and after flocculation. These findings assist in developing an understanding of how oxygen mass transfer characteristics may be affected by activated sludge suspensions, and hence assist in reducing the operational costs of WWTPs.

In the thermoluminescence (TL) detection method for irradiated foods, accurate standards have been developed for detecting irradiated foods. The standard method describes that emission maximum temperature ( T1 i ) and TL ratio for non-heated or non-mixed sample can be in the range of 150–250 °C and more than 0.1, respectively, when it was irradiated food. But when irradiated food is heated up to 200 °C, or mixed up with non-irradiated stuffs, T1 i and TL ratio would not drop in the range. Here we examined the effects of the two processes, heating and mixing with non-irradiated food, on T1 i and G1/G1k ratio (ratio of G1 and average G1 for 1-kGy-irradiated JF2, this value is modeled after TL ratio) using a model consisting of irradiated and non-irradiated geochemical standards of feldspar (JF1, JF2, PF, etc.). T1 i temperatures for irradiated JF1, JF2, and PF ranged from 163 to 175 °C, while those for the non-irradiated JF2 ranged from 253 to 263 °C. T1 i temperatures for 5-kGy-irradiated and preheated JF2 for 10 s, 20 s, and 30 s at 180 °C were 215, 225, and 231 °C, respectively. When JF2 was irradiated from 100 Gy to 5 kGy, the T1 i was almost constant at any doses. G1/G1k ratios at 100, 200, and 500 Gy were 0.15, 0.23, and 0.60, respectively. G1/G1k ratio was proportional to the given dose at the integration temperature ranges. The TS sample, which originated from farm soil in Tanegashima Island, gave the same results as JF2. T1 i s for 5-kGy-irradiated and preheated JF2 for 20 s at 150, 180, and 200 °C were 197, 225, and 246 °C, respectively. Longer and higher preheating resulted in higher T1 i. Longer and higher preheating extremely reduced the G1/G1k ratio, and in some cases the ratio was less than 0.1. This means TL ratio is useless in determination of the standard for irradiated food. Peak temperatures for JF2 in mixture of 5-kGy-irradiated to non-irradiated (1.25–5%) were 261–263 °C (non-irradiated portion, T1 n ) and 177–180 °C (irradiated portion T1 i ). The peak positions are almost the same as those of original components and would not be affected by the mixing ratio. But TL ratio could not be used to determine irradiated food because mixing would reduce it remarkably. Some of the glow curves were simulated by a computer program. In conclusion, T1 i/n is a key factor in an irradiated food determination practice for sample containing feldspar, rather than TL ratio.

The gas–liquid mass transfer in the co-current three-phase fluidized bed with non-Newtonian liquids was studied. Gas hold-ups ( ϕ g ) and volumetric gas–liquid mass transfer coefficients ( k L a ) in a bed of glass beads fluidized in non-Newtonian liquids with gas phase were measured. The gas hold-ups affecting gas–liquid mass transfer rate increased and decreased with increasing particle size and liquid velocity, respectively. The k L a coefficients were evaluated from the dynamic gassing-out method based on the tanks-in-series model describing non-ideal mixing in the three-phase fluidized bed. While an increase in the liquid velocity decreased the k L a coefficient, an increase in the particle size enhanced the k L a coefficient. The increase in purely viscous non-Newtonian flow behaviors reduced gas hold-up and gas–liquid mass transfer rate because of the bubble coalescing nature of the highly viscous non-Newtonian fluid. The theoretical models for ϕ g and k L a in three-phase fluidized beds with non-Newtonian fluids were developed on the basis of the energy dissipation rate. They could reasonably fit the present experimental data.

Lithium/water batteries have attracted considerable attention as high power supply devices because they use high energy density lithium metal as an anode and water as a cathode. In this study, we investigate the use of lithium/water batteries that use a glass–ceramics plate as an electrolyte. A lithium ion conducting glass–ceramics plate has no through-holes and does not exhibit moisture permeation. Such a plate has stable ionic conductivity in water. Lithium/water batteries that used a glass–ceramics plate as an electrolyte had a long and stable discharge for 50 days at room temperature when the lithium metal was prevented from coming into contact with water. Lithium/seawater batteries using a glass–ceramics plate as an electrolyte also operated well in the 10–70 °C temperature range.