A Maximal Correlation Framework for Fair Machine Learning

As machine learning algorithms grow in popularity and diversify to many industries, ethical and legal concerns regarding their fairness have become increasingly relevant. We explore the problem of algorithmic fairness, taking an information–theoretic view. The maximal correlation framework is introduced for expressing fairness constraints and is shown to be capable of being used to derive regularizers that enforce independence and separation-based fairness criteria, which admit optimization algorithms for both discrete and continuous variables that are more computationally efficient than existing algorithms. We show that these algorithms provide smooth performance–fairness tradeoff curves and perform competitively with state-of-the-art methods on both discrete datasets (COMPAS, Adult) and continuous datasets (Communities and Crimes).     

Recent Advances on Metal-Organic Frameworks in the Conversion of Carbon Dioxide

With the development of modern industry,global warming is becoming a challenging issue due to the emissions of large quantities of greenhouse gases,mainly carbon dioxide(CO₂).The conversion of CO₂to useful compounds is considered as an effective and economic way to solve such a climate problem.Metal-organic frameworks(MOFs)are an emerging class of porous crystalline materials that have shown great potential in the conversion of CO₂.The advantages of MOFs in CO 2 conversion lie in their high surface areas,adjustable pore size,and high porosity.More importantly,desirable functional sites can be easily designed and precisely installed to the pore wall of target MOFs by pre-assembly and/or post-synthetic modification(PSM)ways.This review summarizes the recent advances in constructing MOF catalysts for the application in CO₂conversion.We believe that the design and synthesis of MOF catalysts for CO₂conversion can be a promising way to solve the“greenhouse effect”.     

Morphine-induced conditioned place preference in mice: Metabolomic profiling of brain tissue to find “molecular switch” of drug abuse by gas chromatography/mass spectrometry

Conditioned place preference (CPP) is a widely used model to explore the mechanism of context-dependent learning. In this work, we developed a GC–MS method to investigate the metabolites in mice brain which was used to study the mechanism of context-dependent learning associated with rewarding effect of morphine. Metabolites were extracted from brain tissues and derivatized followed by analysis by gas chromatography/mass spectrometry (GC–MS). In total, 69 peaks were identified as known compounds. By a Wilcoxon ran sum test with p value ≤0.05, 21 metabolites were selected and considered as the potential biomarkers of morphine in mice brain. Using principal component analysis (PCA) and receiver-operator characteristic (ROC) curves, a model was constructed with a combination of these 21 metabolic markers. Multivariate statistics of the model yielded separation between the two groups with an area under the curve value of 0.947. Some metabolites were further discussed in detail about their pathway. Results showed that our technique can be successfully applied to profile for biomarkers and in understanding molecular mechanisms of drug abuse.     

Hydrothermal synthesis, structure and quantum chemistry of transition metal complex supported by metal-oxo cluster [(Ni(phen)22 (ξ-Mo8O26)]

A nickel-1,10-phenanthroline complex supported on an octamolybdate, [(Ni(phen)_{2 2}(ξ-Mo₈O₂₆)], has been hydrothermally synthesized with MoO₃, H₂MoO₄, Ni(OAc)₂ 6H₃O and 1,10-phenathroline (1,10-phen) as raw materials. The crystals of the compound belong to monoclinic P2₁/n space group,a = 1.2952(2),b = 1.6659(10),c = 1.3956(12) nm, β =106.273(8)°,V = 2.8906(5) nm³,Z = 2. 5604 observable reflections (I >2σ(I)) were used for structure resolution and refinements to converge to finalR ₁ = 0.0414,wR ₂ = 0.0815. The result of structure determination shows that the compound contains octamolybdate possessing a novel structure type (named as ξ-isomer). The feature of ξ-[Mo₈O₂₆]^4- is that it is composed of Mo₆O₆ ring and two MoO₆ octahedral located at cap positions on opposite faces. The Mo₆O₆ ring contains two octahedral and four trigonal-bipyramidal Mo^VI atoms. Each ξ-[Mo₈O₂₆]^4- unit is bonded with two [Ni(phen)₂]²⁺ through terminal oxygen atoms of octahedral and neighbouring trigonal-bipyramidal Mo atom in the Mo₆O₆ ring. IR and UV-Vis spectra of the compound were measured and its electronic structure was studied by EHMO method.     

Synthesis of CuS@CoS2 Double‐Shelled Nanoboxes with Enhanced Sodium Storage Properties

Metal sulfides have received considerable attention for efficient sodium storage owing to their high capacity and decent redox reversibility. However, the poor rate capability and fast capacity decay greatly hinder their practical application in sodium‐ion batteries. Herein, an elegant multi‐step templating strategy has been developed to rationally synthesize hierarchical double‐shelled nanoboxes with the CoS₂ nanosheet‐constructed outer shell supported on the CuS inner shell. Their structure and composition enable these hierarchical CuS@CoS₂ nanoboxes to show boosted electrochemical properties with high capacity, outstanding rate capability, and long cycle life.     

Bi‐Microporous Metal–Organic Frameworks with Cubane [M4(OH)4] (M=Ni,Co) Clusters and Pore‐Space Partition for Electrocatalytic Methanol Oxidation Reaction

Embedding cubane [M₄(OH)₄] (M=Ni, Co) clusters within the matrix of metal–organic frameworks (MOFs) is a strategy to develop materials with unprecedented synergistic properties. Herein, a new material type based on the pore‐space partition of the cubic primitive minimal‐surface net (MOF‐14‐type) has been realized. CTGU‐15 made from the [Ni₄(OH)₄] cluster not only has very high BET surface area (3537 m² g^?1), but also exhibits bi‐microporous features with well‐defined micropores at 0.86 nm and 1.51 nm. Furthermore, CTGU‐15 is stable even under high pH (0.1 m KOH), making it well suited for methanol oxidation in basic medium. The optimal hybrid catalyst KB&CTGU‐15 (1:2) made from ketjen black (KB) and CTGU‐15 exhibits an outstanding performance with a high mass specific peak current of 527 mA mg^?1 and excellent peak current density (29.8 mA cm^?2) at low potential (0.6 V). The isostructural cobalt structure (CTGU‐16) has also been synthesized, further expanding the application potential of this material type.     

基于透明质酸构筑的药物递送载体及其应用

传统纳米药物控释载体主要通过细胞胞吞作用实现药物递送,其主要过程为被动靶向机制,因此会影响纳米载体在肿瘤组织的富集和治疗效果.近年来生物大分子透明质酸因其优异的水溶性、生物相容性、可降解性和肿瘤靶向性备受科研工作者青睐,已被广泛用于药物控释载体的构筑中,并成为靶向肿瘤治疗纳米载体领域的研究热点.本文根据透明质酸基纳米载...     

Theory of modified thermally stimulated current and direct determination of trap level distribution

In order to investigate the trap level distribution in polymer films, a new method is proposed based on modified thermally stimulated current (TSC) theory and numerical calculation of the TSC measurement. In this method, a new function is defined to weight the contribution of every trap level to the external current. The demarcation energy is used to study the trap emptying process. The modified TSC theory shows that only the electrons with trap levels very close to the demarcation energy can significantly contribute to the external circuit at any instant temperature. Based on this method, the trap level distribution of the DuPont original polyimide film 100HN and nanocomposite polyimide film 100CR are investigated as an application example. The effectiveness of the method is confirmed by the experiments. The experimental results show that the trap level density in the 100CR PI films is about six times larger than that in the 100HN PI films through the investigated trap level ranges 06–1.3 eV. The increased traps in 100CR should be introduced by nanofillers, probably come from the interfaces formed between nanofillers and the polymer matrix.     

Targeting Upconversion Nanoprobes for Magnetic Resonance Imaging of Early Colon Cancer

Colon cancer (CC) is one of the most common intestinal malignancies and is difficult to detect in its early stage by magnetic resonance imaging (MRI) with currently used contrast agents (CAs). The development of targeted CAs contributes to the early diagnosis of CC and thereby enables early intervention and timely therapy. Considering the outstanding performance of upconversion nanoprobes (UCNPs) in high‐performance MR and fluorescence imaging, a new type of nanoprobes with considerably enhanced imaging performance is developed herein. Carcinoembryonic antigen (CEA) antibody is conjugated onto the surface of UCNPs to achieve the targeted imaging of early CC tumors, which overexpress CEA. Both toxicity tests and histological/hematological examinations demonstrate the excellent biocompatibility of these CC‐targeting nanoprobes, which possess great potential for clinical application in the early diagnosis of CC.     

Bounds on an anomalous dijet resonance in W+jets production in pp collisions at √s = 1.96 TeV

We present a study of the dijet invariant mass spectrum in events with two jets produced in association with a W boson in data corresponding to an integrated luminosity of 4.3 fb(-1) collected with the D0 detector at √s = 1.96 TeV. We find no evidence for anomalous resonant dijet production and derive upper limits on the production cross section of an anomalous dijet resonance recently reported by the CDF Collaboration, investigating the range of dijet invariant mass from 110 to 170 GeV/c(2). The probability of the D0 data being consistent with the presence of a dijet resonance with 4 pb production cross section at 145 GeV/c(2) is 8×10(-6).