Neural networks can learn to represent and manipulate numerical information, but they seldom generalize well outside of the range of numerical values encountered during training. To encourage more systematic numerical extrapolation, we propose an architecture that represents numerical quantities as linear activations which are manipulated using primitive arithmetic operators, controlled by learned gates. We call this module a neural arithmetic logic unit (NALU), by analogy to the arithmetic logic unit in traditional processors. Experiments show that NALU-enhanced neural networks can learn to track time, perform arithmetic over images of numbers, translate numerical language into real-valued scalars, execute computer code, and count objects in images. In contrast to conventional architectures, we obtain substantially better generalization both inside and outside of the range of numerical values encountered during training, often extrapolating orders of magnitude beyond trained numerical ranges.


    Access

    Download


    Export, share and cite



    Title :

    Neural arithmetic logic units


    Contributors:
    Trask, A (author) / Hill, F (author) / Reed, SE (author) / Rae, J (author) / Dyer, C (author) / Blunsom, P (author)

    Publication date :

    2019-08-16


    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English


    Classification :

    DDC:    629



    Pipelining of arithmetic units

    Hallin, Thomas G. | TIBKAT | 1970


    Modelling Truth Tables, Boolean Logic and Arithmetic Operations

    Howden, P. F. / Modelling and Simulation Society of Australia | British Library Conference Proceedings | 1995


    ARITHMETIC LOGIC DEVICE, CAMERA DEVICE, VEHICLE AND CALIBRATION METHOD

    INOUE SHUSHIN / KOISHI TOMOFUMI | European Patent Office | 2020

    Free access


    Design and Implementation of a Reversible Logic based 8-Bit Arithmetic and Logic Unit

    Arunachalam, K. / Perumalsamy, M. / Sundaram, C.K. et al. | British Library Online Contents | 2014