Temporal State Machines: Using Temporal Memory to Stitch Time-based Graph Computations

Race logic, an arrival-time-coded logic family, has demonstrated energy and performance improvements for applications ranging from dynamic programming to machine learning. However, the various ad hoc mappings of algorithms into hardware rely on researcher ingenuity and result in custom architectures that are difficult to systematize. We propose to associate race logic with the mathematical field of tropical algebra, enabling a more methodical approach toward building temporal circuits. This association between the mathematical primitives of tropical algebra and generalized race logic computations guides the design of temporally coded tropical circuits. It also serves as a framework for expressing high-level timing-based algorithms. This abstraction, when combined with temporal memory, allows for the systematic exploration of race logic-based temporal architectures by making it possible to partition feed-forward computations into stages and organize them into a state machine. We leverage analog memristor-based temporal memories to design such a state machine that operates purely on time-coded wavefronts. We implement a version of Dijkstra's algorithm to evaluate this temporal state machine. This demonstration shows the promise of expanding the expressibility of temporal computing to enable it to deliver significant energy and throughput advantages.

Chiral-Bubble-Induced Topological Hall Effect in Ferromagnetic Topological Insulator Heterostructures

We report compelling evidence of an emergent topological Hall effect (THE) from chiral bubbles in a two-dimensional uniaxial ferromagnet, V-doped Sb₂Te₃ heterostructure. The sign of THE signal is determined by the net curvature of domain walls in different domain configurations, and the strength of THE signal is correlated with the density of nucleation or pinned bubble domains. The experimental results are in good agreement with the integrated linear transport and Monte Carlo simulations, corroborating the emergent gauge field at chiral magnetic bubbles. Our findings not only reveal a general mechanism of THE in two-dimensional ferromagnets but also pave the way for the creation and manipulation of topological spin textures for spintronic applications.

Roadmap on emerging hardware and technology for machine learning

Recent progress in artificial intelligence is largely attributed to the rapid development of machine learning, especially in the algorithm and neural network models. However, it is the performance of the hardware, in particular the energy efficiency of a computing system that sets the fundamental limit of the capability of machine learning. Data-centric computing requires a revolution in hardware systems, since traditional digital computers based on transistors and the von Neumann architecture were not purposely designed for neuromorphic computing. A hardware platform based on emerging devices and new architecture is the hope for future computing with dramatically improved throughput and energy efficiency. Building such a system, nevertheless, faces a number of challenges, ranging from materials selection, device optimization, circuit fabrication and system integration, to name a few. The aim of this Roadmap is to present a snapshot of emerging hardware technologies that are potentially beneficial for machine learning, providing the Nanotechnology readers with a perspective of challenges and opportunities in this burgeoning field.

Energy-efficient stochastic computing with superparamagnetic tunnel junctions

Superparamagnetic tunnel junctions (SMTJs) have emerged as a competitive, realistic nanotechnology to support novel forms of stochastic computation in CMOS-compatible platforms. One of their applications is to generate random bitstreams suitable for use in stochastic computing implementations. We describe a method for digitally programmable bitstream generation based on pre-charge sense amplifiers. This generator is significantly more energy efficient than SMTJ-based bitstream generators that tune probabilities with spin currents and a factor of two more efficient than related CMOS-based implementations. The true randomness of this bitstream generator allows us to use them as the fundamental units of a novel neural network architecture. To take advantage of the potential savings, we codesign the algorithm with the circuit, rather than directly transcribing a classical neural network into hardware. The flexibility of the neural network mathematics allows us to adapt the network to the explicitly energy efficient choices we make at the device level. The result is a convolutional neural network design operating at ≈ 150 nJ per inference with 97 % performance on MNIST-a factor of 1.4 to 7.7 improvement in energy efficiency over comparable proposals in the recent literature.

Spin chirality fluctuation in two-dimensional ferromagnets with perpendicular magnetic anisotropy

Non-coplanar spin textures with scalar spin chirality can generate an effective magnetic field that deflects the motion of charge carriers, resulting in a topological Hall effect (THE)^1-3. However, spin chirality fluctuations in two-dimensional ferromagnets with perpendicular magnetic anisotropy have not been considered so far. Here, we report evidence of spin chirality fluctuations by probing the THE above the Curie temperature in two different ferromagnetic ultra-thin films, SrRuO₃ and V-doped Sb₂Te₃. The temperature, magnetic field, thickness and carrier-type dependence of the THE signal, along with Monte Carlo simulations, suggest that spin chirality fluctuations are a common phenomenon in two-dimensional ferromagnets with perpendicular magnetic anisotropy. Our results open a path for exploring spin chirality with topological Hall transport in two-dimensional magnets and beyond^4-7.

Streaming Batch Eigenupdates for Hardware Neural Networks

Neural networks based on nanodevices, such as metal oxide memristors, phase change memories, and flash memory cells, have generated considerable interest for their increased energy efficiency and density in comparison to graphics processing units (GPUs) and central processing units (CPUs). Though immense acceleration of the training process can be achieved by leveraging the fact that the time complexity of training does not scale with the network size, it is limited by the space complexity of stochastic gradient descent, which grows quadratically. The main objective of this work is to reduce this space complexity by using low-rank approximations of stochastic gradient descent. This low spatial complexity combined with streaming methods allows for significant reductions in memory and compute overhead, opening the door for improvements in area, time and energy efficiency of training. We refer to this algorithm and architecture to implement it as the streaming batch eigenupdate (SBE) approach.

Abstract P1-05-04: Progestin receptor content of breast cancer associated with expression of gene subsets for peptide/protein hormones and cognate receptors in LCM-procured cells that impact clinical outcomes

Assessing breast cancer prognosis and treatment selection was first enhanced by discovery that estrogen (ER) and progestin receptor (PR) proteins were clinically relevant biomarkers. Later reports suggested expression of certain protein hormones in breast cancer cells appear to be related to clinical behavior. We deciphered clinically relevant relationships between PR content in tissue biopsies and relative expression levels of genes directing synthesis of peptide/protein hormones and their cognate receptors in LCM-procured breast cancer cells. This global approach revealed unique relationships between both PR+ or PR- carcinomas and small subsets of genes for these peptide/protein hormones and cognate receptors as independent predictors of risk of recurrence.

Methods: Expression of genes for 61 peptide/protein hormones and 81 cognate receptor proteins were measured by microarray analyses of LCM-procured carcinoma cells from de-identified primary breast carcinoma biopsies. Using an IRB-approved biorepository and database, previously total RNA was extracted from carcinoma cells to determine expression levels of ˜22,000 genes. Univariable and multivariable Cox regressions with interaction of each hormone

eceptor gene, individually or paired, and use of LASSO were determined with relative gene expression values of each protein ligand and its cognate receptor. PR content and ligand binding affinity of each carcinoma were quantified by FDA approved assays.

Results: Using de-identified clinical outcomes that extended up to 12 years, univariable Cox regression analyses of 142 candidates revealed AVPR1A, AVPR2, CALCR, CRH, LHB, POMC, SCT, SST, SSTR1 and TMSB10 independently predicted PFS or OS. Violin plots identified candidate genes associated with PR content. Multivariable analyses of relative gene expression of 115 hormone-receptor pairs showed IAPP-CALCR, RLN2-RXFP1, GHRH-GHRHR, CGA-TSHR, EDN1-ENDRA and POMC-MC5R exhibited statistically significant interaction for predicting OS among 145 PR+ cancers. Four gene pairs (HCRT-HCRTR2, CRH-CRHR1, HCRT-HCRTR1 and CORT-SSTR4) were associated with OS among 101 PR- lesions. Using LASSO, PR+ lesions expressing either (CGA & SSTR2) or (CGA) predicted OS and PFS, respectively. Similarly, for PR- cancers, gene subsets (GRP, TMSB15, VIP2) or (AGT, GH1, GRP, TMSB15A) predicted OS and PFS, respectively. Three of four signatures were externally validated with SurvExpress.

Conclusion: Different gene expression profiles for protein hormones and cognate receptors were identified in PR+ or PR- carcinomas at time of diagnosis that predict PFS and OS regardless of treatment. Combining PR content with gene expression of LCM-procured cells is likely to provide insight into alternative treatments whereby standard of care performed poorly or to identify genes that correlate with cancer progression regardless of treatment modality. Collectively, results suggest that many primary breast cancers exhibit considerable endocrine autonomy for controlling disease progression, supporting investigation of protein products of gene candidates in isolated populations of breast carcinoma cells to develop novel biomarker assays.

Citation Format: Daniels MW, Brock GN, James WL. Progestin receptor content of breast cancer associated with expression of gene subsets for peptide/protein hormones and cognate receptors in LCM-procured cells that impact clinical outcomes [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P1-05-04.

Abstract P1-01-12: Expression of autophagy related genes impacts clinical outcomes of human breast carcinoma and is associated with estrogen and progestin receptor status

The dual role of autophagy in breast cancer initiation, progression and responsiveness to various therapies is the focus of extensive studies. Our goal is to assess the relationship of expression of certain autophagy related genes in primary breast carcinomas to predict risk of recurrence. Our hypothesis includes the caveat that unique gene expression subsets will be deciphered by utilizing Laser Capture Microdissected breast carcinoma cells from tissue biopsies containing many cell types.

Methods: Comprehensive analyses of de-identified biomarker results, characteristics of 247 breast carcinoma specimens and clinical outcomes were performed by univariable Cox regressions, Kaplan Meier plots and LASSO with R software version 3.2.5. Microarray analyses were performed on RNA isolated from LCM-procured carcinoma cells to identify gene signatures associated with breast cancer behavior.

Results: Expression levels of 22 autophagy related genes analyzed by univariable Cox regression revealed that RB1CC1, KEAP1, ATG7, RUBCN, NOD2, HMOX1, BECN1, PIK3R4 were significant for predicting Progression Free Survival (PFS) at the discovery level of the adjusted p-value < 0.3. Of these, RB1CC1, KEAP1, & ATG7 were significant for Overall Survival (OS) without regard to ER/PR status. Using Kaplan Meier analyses, expression levels of each of 10 candidate genes predicted PFS of which expression of 6 of these genes also predicted OS using a median split cutoff without regard to ER/PR status. Applying LASSO computations without regard to ER/PR status of the primary breast cancer, a clinically relevant gene expression profile consisting of ATG7, BCL2, HMOX1, KEAP1, NOD2, PTEN, RB1CC1, and ULK4 predicted PFS. Collectively, expression of these 8 genes with TP53 and RUBCN predicted OS without regard to ER/PR status. Violin plots of relative gene expression of each of the 22 candidate genes known to be associated with autophagy pathways revealed that ATG7, ATG12, BCL2 and BECN1 were elevated in ER+ lesions while BCL2 and BECN1 were also elevated in PR+ breast carcinomas. Expression levels of 16 of the 22 autophagy related genes examined by univariable Cox regression were related to either ER or PR status of the primary lesion and predicted either PFS or OS. Refinement of clinically relevant gene subsets was accomplished by LASSO calculations in which either the ER or PR protein status of the primary breast carcinoma was considered. Of the genes in the molecular signatures derived, expression levels in breast carcinomas separated by either ER or PR status of the lesion were tested by Kaplan Meier analyses to assess relationships to PFS and OS.

Conclusions: Using gene expression results derived from microarray analyses of LCM-procured breast carcinoma cells of primary lesions, subsets of autophagy related genes were identified that predict a patient's risk of recurrence and overall survival. Expression of a number of candidate genes appears to be related to either/both the ER or PR protein status of the primary lesion. Collectively, results suggest that expression of certain autophagy related genes may serve as biomarkers for assessing prognosis of breast carcinoma thus impacting clinical management of breast cancer.

Citation Format: Wittliff JL, Hameed ZR, Daniels MW, Cheng A. Expression of autophagy related genes impacts clinical outcomes of human breast carcinoma and is associated with estrogen and progestin receptor status [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P1-01-12.

Abstract P3-05-11: Clinical outcomes of estrogen-receptor negative breast carcinoma are associated with protein hormone-cognate receptor gene expression

Background: Since certain protein hormones appear to be associated with progression of breast cancer cells in vitro, we employed a global approach by evaluating relationships between genes for these hormones and their cognate receptors in human breast cancer biopsies as independent predictors of risk of recurrence. Our goal is to derive clinically relevant molecular signatures correlating various ER/PR subtypes.

Methods: Microarray analyses of LCM-procured carcinoma cells from 247 de-identified biopsies determined expression of genes for 61 protein hormones (the ligands) and 81 cognate receptors. Total RNA was extracted, purified and amplified from cells to determine expression of 22,000 genes. Univariable and multivariable Cox regressions were determined using expression levels of each hormone/receptor gene. Kaplan-Meier plots were constructed with an adjusted p-value <0.30 (discovery cutoff) for each gene candidates with/without ER status of the lesion. Cox regression analyses with/without interaction models deciphered candidates of hormone-receptor complexes that predicted risk of recurrence as well as overall survival using clinical follow-up that extended as much as 12 years. LASSO was used to derive gene subsets to predict overall survival (OS) and progression free survival (PFS) within ER/PR subtypes.

Results: Expression levels of 7 genes for protein hormones and 10 receptor genes were significant for OS at an adjusted p-value of <0.30 while expression of 15 hormone and 19 receptor genes were significant for PFS without regard to ER/PR subtypes. When expression levels were considered with ER status only, 3 genes for ER+ lesions and 6 genes for ER- lesions were identified that predicted OS/PFS. Categorization of cancers according to combined ER and PR status revealed one gene for ER+/PR+, 3 genes for ER+/PR- and 2 genes for ER-/PR- associated with PFS/OS. After stratifying gene expression into above/below median for the linear predictor, Kaplan-Meier plots revealed patient groups with each signature for ER+/PR+, ER-/PR- and ER- lesions that exhibited significantly different survival. Molecular signatures comprising 16 genes for ER+/PR+ carcinomas, 10 genes for ER-/PR- lesions and 6 genes for ER- carcinomas predicted clinical outcomes with C-indices of 0.79, 0.76 and 0.71, respectively.

Conclusion: We revealed many breast carcinomas synthesize mRNA species for various protein hormones and their cognate receptors by determining gene expression directly on pure populations of these cells procured by LCM. We report a novel ten-gene ER-/PR- signature containing four genes in common with that of a six-gene ER- only signature that predicts breast cancer recurrence. Collectively, results of mRNA expression suggest that often breast carcinomas exhibit substantial elements of endocrine autonomy for regulating progression, warranting investigation of protein products of gene candidates in isolated populations of breast carcinoma cells.

Citation Format: Daniels MW, Wittliff JL, Brock GN. Clinical outcomes of estrogen-receptor negative breast carcinoma are associated with protein hormone-cognate receptor gene expression [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P3-05-11.

Abstract P3-04-06: Expression of genes for peptide/protein hormones and their receptors in breast carcinomas as biomarkers predicting risk of recurrence

Background: Several reports have suggested expression of certain peptide/protein hormones in breast cancer cells appear to be related to clinical behavior. We have taken a global approach by evaluating relationships between genes for these hormones and their cognate receptor proteins as independent predictors of risk of recurrence.

Methods: Expression of genes for 55 peptide/protein hormones (the ligands) and 73 of their cognate receptor proteins were measured by microarray analyses of LCM-procured carcinoma cells from 247 breast carcinoma biopsies. Using an IRB-approved biorepository and comprehensive database, total RNA was extracted from carcinoma cells to determine expression levels of 22,000 genes. Univariate and multivariate Cox regressions with interaction were determined using expression values of each ligand and its cognate receptor with an interaction term.

Results: When pairs of hormone ligand and cognate receptor genes were evaluated by multivariate Cox regression with interaction, the following sets of genes were identified that predicted risk of breast cancer recurrence (INS/IGF2R, HAMP/SLC40A1, POMC/MC4R, GH1/GHR and VIP/VIPR2) and OS (CORT/SSTR5, VIP/VIPR2 and GHRH/GHRHR, based on unadjusted p-value for interaction term < 0.05). Since expression in situ of both hormone and receptor are necessary to elicit endocrine action, a unique alternative approach using the minimum expression value between ligand and receptor for each patient was applied. These results revealed that the complex of HAMP/SLC40A1 showed significance in both the interaction and minimum models, and was further investigated by splitting the ligand and receptor into low (below 1st quartile) and high (above 1st quartile) expression groups. The group consisting of high expression for both ligand and receptor was contrasted with the other three groups (low expression for ligand, receptor, or both). The higher expression group exhibited a better DFS (hazard ratio (HR) = 0.56 95% CI 0.37-0.84, p=0.004) and OS (HR = 0.59 95% CI 0.37-0.93, p=0.022).

Conclusion: As a result of determining gene expression directly on pure populations of breast carcinoma cells procured by LCM, we have demonstrated the many lesions synthesize mRNA species for a wide variety of peptide and protein hormones as well as for their cognate receptor proteins. Using clinical follow-up that extended as much as 12 years, univariate and multivariate Cox regression analyses with and without interaction models revealed a number of noteworthy candidates of hormone-receptor complexes that predicted risk of breast cancer recurrence as well as overall survival. Collectively, our results suggests that many breast carcinomas exhibit considerable endocrine autonomy for controlling progression, which warrants investigation of the protein products of the gene candidates in isolated populations of breast carcinoma cells.

Citation Format: Wittliff JL, Daniels MW, Brock GN. Expression of genes for peptide/protein hormones and their receptors in breast carcinomas as biomarkers predicting risk of recurrence. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P3-04-06.

Abstract P6-03-14: Expression of genes for aromatase inhibitor targets to discriminate invasive lobular from invasive ductal carcinomas of the breast using LCM-procured cells to complement endocrine biomarkers

Background: The BIG 1-98 trial and later the ABCSG 8 study reported that patients with invasive lobular carcinoma (ILC) exhibited better response to aromatase inhibitors (AIs) compared to those with invasive ductal carcinoma (IDC). Aromatase cytochrome P450 (CYP19) synthesizes estrogen from androgens and is the target of AI's. CYP19 substrates are generated by upstream enzymes including estrone sulfatase (SULT1E1) and 3b-hydroxysteroid dehydrogenase type 1 (HSD3B1). Enzymes of the aromatase pathway have been reported to be expressed in intact tissue biopsies of breast cancer. To learn more about the pathogenic mechanisms that may underlie the survival benefit of ILCs treated with AIs, we analyzed expression levels of key enzymes related to the aromatase pathway in ILC and IDC. Unlike previous studies, we determined gene expression levels directly on pure populations of carcinoma cells procured by laser capture microdissection, eliminating the contribution of non-cancerous cells.

Methods: Using an IRB-approved biorepository and database, total RNA was extracted from carcinoma cells of 247 de-identified biopsies to perform microarray analyses of 22,000 genes. Of the 247 samples, 16 were ILC, 13 were low grade IDC, 55 were intermediate grade IDC and 85 were high grade IDC, and 107 of these were hormone receptor positive. CYP19, HSD3B1 and SULT1E1 expression was directly detected in LCM-procured breast carcinoma cells of ILC and of IDC. Expression of other genes generally associated with the aromatase pathway, e.g., NADPH-cytochrome P450 reductase (POR), ATP-binding cassette gene (ABCG2), catechol-o-methyltransferase (COMT) and uridine-5'-diphosphate glucuronosyltransferase (UGT1A3 & UGT1A9) as well as HSD17B2 were assessed with LCM-procured cells. Estrogen receptor (ER) and progesterone receptor (PR) proteins were quantified by radio-ligand binding and EIA, and gene expression was validated by qPCR.

Results: Univariate Cox regression analyses indicated that ABCG2, HSD17B2and UGTA3 independently predicted disease free and/or overall survival of breast carcinomas. We found that CYP19 expression in carcinoma cells, as well as SULT1E1, COMT, POR, HSD17B2 and UTG1A3 expression, decreased as either ER or PR protein increased. HSD3B1 appeared to be over-expressed in ILC compared to IDC, however this difference did not approach statistical significance, likely due to the small sample size. No differences were seen in expression levels of CYP19 and SULT1E1 between ILC and IDC.

Conclusions: An inverse relationship between CYP19 and ER and PR expression levels was observed and suggests that synthesis of estrogens by breast cancer cells in situ plays a significant role in defining tumor biology. Our results also indicate overexpression of HSD3B1 in ILC, although not statistically significant. This finding suggests that HSD3B1 may be a key contributor to the increased benefit of AI therapy seen in ILC. Collectively our results suggest a comprehensive study is warranted to ascertain the molecular basis for differences in expression of genes directing estrogen synthesis in situ in relationship to AI therapeutic responses of histologic subtypes of breast carcinomas.

Citation Format: Sanders MAG, Daniels MW, Wittliff JL. Expression of genes for aromatase inhibitor targets to discriminate invasive lobular from invasive ductal carcinomas of the breast using LCM-procured cells to complement endocrine biomarkers. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P6-03-14.

Reactions of a Trifunctional Silane Coupling Agent in the Presence of Colloidal Silica Sols in Polar Media

Reactions of 3-glycidoxypropyltrimethoxysilane (GPS) in a water-rich environment with silica colloids were studied. (29)Si NMR spectroscopy was used to monitor quantitatively the variation in GPS hydrolysis, condensation, and adsorption with pH and water concentration. The results show the competition between adsorption on the silica surface and condensation polymerization of GPS in the bulk liquid and are used to identify conditions favorable to surface modification. The presence of colloidal silica was shown to increase the apparent GPS dimerization rate, suggesting either surface catalysis or an increased local reactant concentration in the electrical double layer around the silica colloids. Copyright 1999 Academic Press.

Silane Adsorption Behavior, Microstructure, and Properties of Glycidoxypropyltrimethoxysilane-Modified Colloidal Silica Coatings

The adsorption of 3-glycidoxypropyltrimethoxysilane (GPS) onto silica colloids (20 nm diameter) suspended in an isopropanol/water solution was studied and related to the microstructure and properties of coatings prepared from the suspensions. 29Si NMR results indicate an adsorption limit of approximately 2.2 GPS monomer or repeat units per nm2 of silica surface after 24 h in suspension. For coatings dried at 100 degreesC, FTIR data show that surface silanols are further consumed as GPS is added beyond the adsorption limit in the suspension. The increased adsorption after drying may be related to enhanced reaction at the higher temperature and to drying-induced deposition onto the surfaces. Coating microstructures consist of randomly packed silica colloids with varying amounts of siloxane polymer filling the pores, depending on the amount of silane added. The critical cracking thickness of the coatings increased dramatically when the amount of silane added was sufficient to fill the porespace. Abrasion resistance of the coatings improved gradually with GPS content, but deteriorated when the GPS content was far in excess of that needed to fill the pores. Hardness measured by nanoindentation showed a similar trend as abrasion resistance. Copyright 1998 Academic Press.