A Memristor-Based Learning Engine for Synaptic Trace-Based Online Learning.

The memristor has been extensively used to facilitate the synaptic online learning of brain-inspired spiking neural networks (SNNs). However, the current memristor-based work can not support the widely used yet sophisticated trace-based learning rules, including the trace-based Spike-Timing-Dependent Plasticity (STDP) and the Bayesian Confidence Propagation Neural Network (BCPNN) learning rules. This paper proposes a learning engine to implement trace-based online learning, consisting of memristor-based blocks and analog computing blocks. The memristor is used to mimic the synaptic trace dynamics by exploiting the nonlinear physical property of the device. The analog computing blocks are used for the addition, multiplication, logarithmic and integral operations. By organizing these building blocks, a reconfigurable learning engine is architected and realized to simulate the STDP and BCPNN online learning rules, using memristors and 180 nm analog CMOS technology. The results show that the proposed learning engine can achieve energy consumption of 10.61 pJ and 51.49 pJ per synaptic update for the STDP and BCPNN learning rules, respectively, with a 147.03× and 93.61× reduction compared to the 180 nm ASIC counterparts, and also a 9.39× and 5.63× reduction compared to the 40 nm ASIC counterparts. Compared with the state-of-the-art work of Loihi and eBrainII, the learning engine can reduce the energy per synaptic update by 11.31× and 13.13× for trace-based STDP and BCPNN learning rules, respectively.

Mapping the BCPNN Learning Rule to a Memristor Model.

The Bayesian Confidence Propagation Neural Network (BCPNN) has been implemented in a way that allows mapping to neural and synaptic processes in the human cortexandhas been used extensively in detailed spiking models of cortical associative memory function and recently also for machine learning applications. In conventional digital implementations of BCPNN, the von Neumann bottleneck is a major challenge with synaptic storage and access to it as the dominant cost. The memristor is a non-volatile device ideal for artificial synapses that fuses computation and storage and thus fundamentally overcomes the von Neumann bottleneck. While the implementation of other neural networks like Spiking Neural Network (SNN) and even Convolutional Neural Network (CNN) on memristor has been studied, the implementation of BCPNN has not. In this paper, the BCPNN learning rule is mapped to a memristor model and implemented with a memristor-based architecture. The implementation of the BCPNN learning rule is a mixed-signal design with the main computation and storage happening in the analog domain. In particular, the nonlinear dopant drift phenomenon of the memristor is exploited to simulate the exponential decay of the synaptic state variables in the BCPNN learning rule. The consistency between the memristor-based solution and the BCPNN learning rule is simulated and verified in Matlab, with a correlation coefficient as high as 0.99. The analog circuit is designed and implemented in the SPICE simulation environment, demonstrating a good emulation effect for the BCPNN learning rule with a correlation coefficient as high as 0.98. This work focuses on demonstrating the feasibility of mapping the BCPNN learning rule to in-circuit computation in memristor. The feasibility of the memristor-based implementation is evaluated and validated in the paper, to pave the way for a more efficient BCPNN implementation, toward a real-time brain emulation engine.

The development of ß-selective glycosylation reactions with benzyl substituted 2-deoxy-1,4-dithio-D-erythro-pentofuranosides: enabling practical multi-gram syntheses of 4'-Thio-2'-deoxycytidine (T-dCyd) and 5-aza-4'-thio-2'-deoxycytidine (aza-T-dCyd) to support clinical development.

The lack of effective methods to perform direct ß-selective glycosylation reactions with 2-deoxy-1,4-dithio-D-erythro-pentofuranosides has long been a significant stumbling block for the multi-gram synthesis of 4'-thio-2'-deoxy nucleosides. In addition, previously reported methods for the preparation of appropriately substituted 2-deoxy-1,4-dithio-D-erythro-pentofuranosides have proven problematic for large scale synthesis. To address these issues, herein we describe the modification and optimization of previously reported methods to allow for the convenient large scale synthesis of benzyl substituted 2-deoxy-1,4-dithio-D-erythro-pentofuranosides. Furthermore, we describe the development of reaction conditions for ß-selective glycosylation reactions of benzyl substituted 2-deoxy-1,4-dithio-D-erythro-pentofuranosides with both N4-benzoylcytosine and 5-aza-cytosine to enable the practical multi-gram syntheses of the clinical candidates 4'-thio-2'-deoxycytidine (T-dCyd) and 5-aza-4'-thio-2'-deoxycytidine (aza-T-dCyd). Taken together, these new synthetic developments have made possible the preclinical and early clinical development of these important anticancer agents at the National Cancer Institute.

Atropisomerism of a monosubstituted perfluoro[2.2]paracyclophane. A combined synthetic, kinetic, spectroscopic and computational study.

The product of S(N)Ar addition of the enolate of ethyl acetoacetate to perfluoro[2.2]para-cyclophane exists entirely as its enol tautomer 5. This enol exhibits two NMR signals for its enolic proton, and these signals were shown to derive from the presence of two, equal energy conformations that were observable as distinct, stable conformations at room temperature, but which when heated, interconverted with an energy barrier of 23.5 kcal mol(-1). These atropisomers were characterized by NMR, with details of this analysis being provided. Computational work corroborated the NMR conclusions, and provided additional insight into all structural, thermodynamic and kinetic results. Enol product 5 was cyclized, under basic conditions, to form a benzofuran product 6. Its structure was confirmed by NMR, with further structural and mechanistic insights being provided by calculations.

19F chemical shifts, coupling constants and conformational preferences in monosubstituted perfluoroparacyclophanes.

In the process of studying the chemistry of perfluoro[2.2]paracyclophanes (PFPCs), a novel class of compounds, it became necessary to identify some disubstituted products. To achieve this goal, we characterize in this work some monosubstituted PFPCs, identifying their (19)F-(19) F coupling patterns, and establishing a methodology for the assignment of their (19)F chemical shifts. The pattern of coupling constants indicates a skewed geometry in which the upper deck moves towards or away from the substituent, depending on the substituent electron-donor character and size. Quantum chemical calculations, performed at the HF/6-311 + G(d,p)//B3LYP/EPR-III level of theory, confirmed the conformations inferred from coupling constants and reproduced well the values of the couplings. Transmission mechanisms for the FC term of four- and five-bond (19)F-(19) F couplings are discussed in detail. Understanding the conformational preferences of PFPCs and how they are reflected by the coupling constants facilitates the assignment of (19)F chemical shifts in monosubstituted PFPCs and the identification of the disubstituted products.

Reactions of nucleophiles with perfluoro[2.2]paracyclophane.

The aromatic rings of perfluoro[2.2]paracyclophane are extremely reactive with respect to nucleophilic substitution reactions. This paper emphasizes products of monosubstitution by hydroxide, alkoxide, thiolate, enolate, and amine nucleophiles. All reactions appear to proceed via S(N)Ar mechanisms; reactivity issues are discussed including the effect of substituents on reactivity and regiochemistry of substitution.

Synthesis of perfluoro[2.2]paracyclophane.

A synthesis of perfluoro[2.2]paracyclophane has been sought ever since the partially fluorinated octafluoro[2.2]paracyclophane (AF4) was prepared and its chemistry studied. This compound has now been prepared in 39% yield from the precursor, 1,4-bis(chlorodifluoromethyl)-2,3,5,6-tetrafluorobenzene by its reaction with Zn when heated in acetonitrile at 100 degrees C. Two preparations of the precursor, first from 1,4-dicyano-2,3,5,6-tetrachlorobenzene and an improved method beginning from 1,2,4,5-tetrachlorobenzene, are also described as are key comparisons to our related synthesis of AF4.

Influence of various central moieties on the hypolipidemic properties of long hydrocarbon chain diols and diacids.

A series of long (11-15) hydrocarbon chain diols and diacids with various central functional groups and terminal gem-dimethyl or -methyl/aryl substituents was synthesized and evaluated in both in vivo and in vitro assays for its potential to favorably alter lipid disorders including metabolic syndrome. Compounds were assessed for their effects on the de novo incorporation of radiolabeled acetate into lipids in primary cultures of rat hepatocytes, as well as for their effects on lipid and glycemic variables in obese female Zucker fatty rats, Crl:(ZUC)-faBR. The most active compounds were hydroxyl-substituted symmetrical diacids and diols with a 13-atom chain and terminal gem-dimethyl substituents. Furthermore, biological activity was enhanced by central substitution with O, C=O, S, S=O compared to the methylene analogues and was diminished for compounds with central functional groups such as carbamate, ester, urea, acetylmethylene, and hydroxymethylene.