Electric-field Inputs for Molecular Quantum-dot Cellular Automata Circuits
Authors:
Enrique P. Blair
Abstract:
Quantum-dot cellular automata (QCA) is a low-power, non-von-Neumann, general-purpose paradigm for classical computing using transistor-free logic. An elementary QCA device called a "cell" is made from a system of coupled quantum dots with a few mobile charges. The cell's charge configuration encodes a bit, and quantum charge tunneling within a cell enables device switching. Arrays of cells network…
▽ More
Quantum-dot cellular automata (QCA) is a low-power, non-von-Neumann, general-purpose paradigm for classical computing using transistor-free logic. An elementary QCA device called a "cell" is made from a system of coupled quantum dots with a few mobile charges. The cell's charge configuration encodes a bit, and quantum charge tunneling within a cell enables device switching. Arrays of cells networked locally via the electrostatic field form QCA circuits, which mix logic, memory and interconnect. A molecular QCA implementation promises ultra-high device densities, high switching speeds, and room-temperature operation. We propose a novel approach to the technical challenge of transducing bits from larger conventional devices to nanoscale QCA molecules. This signal transduction begins with lithographically-formed electrodes placed on the device plane. A voltage applied across these electrodes establishes an in-plane electric field, which selects a bit packet on a large QCA input circuit. A typical QCA binary wire may be used to transmit a smaller bit packet of a size more suitable for processing from this input to other QCA circuitry. In contrast to previous concepts for bit inputs to molecular QCA, this approach requires neither special QCA cells with fixed states nor nanoelectrodes which establish fields with single-electron specificity. A brief overview of the QCA paradigm is given. Proof-of-principle simulation results are shown, demonstrating the input concept in circuits made from two-dot QCA cells. Importantly, this concept for bit inputs to molecular QCA may enable solutions to or provide insights into other challenges to the realization of molecular QCA, such as the demonstration of molecular device switching, the read-out of molecular QCA states, and the layout of molecular QCA circuits.
△ Less
Submitted 14 August, 2018; v1 submitted 10 May, 2018;
originally announced May 2018.
Automated Generation of Multilingual Clusters for the Evaluation of Distributed Representations
Authors:
Philip Blair,
Yuval Merhav,
Joel Barry
Abstract:
We propose a language-agnostic way of automatically generating sets of semantically similar clusters of entities along with sets of "outlier" elements, which may then be used to perform an intrinsic evaluation of word embeddings in the outlier detection task. We used our methodology to create a gold-standard dataset, which we call WikiSem500, and evaluated multiple state-of-the-art embeddings. The…
▽ More
We propose a language-agnostic way of automatically generating sets of semantically similar clusters of entities along with sets of "outlier" elements, which may then be used to perform an intrinsic evaluation of word embeddings in the outlier detection task. We used our methodology to create a gold-standard dataset, which we call WikiSem500, and evaluated multiple state-of-the-art embeddings. The results show a correlation between performance on this dataset and performance on sentiment analysis.
△ Less
Submitted 5 April, 2017; v1 submitted 4 November, 2016;
originally announced November 2016.
