Unsupervised Learning in a Framework of Information Compression by Multiple Alignment, Unification and Search

Syntax, Parsing and Production of Natural Language in a Framework of Information Compression by Multiple Alignment, Unification and Search

This article introduces the idea that "information compression by multiple alignment, unification and search" (ICMAUS) provides a framework within which natural language syntax may be represented in a simple format and the parsing and production of natural language may be performed in a transparent manner. The ICMAUS concepts are embodied in a software model, SP61. The organisation and operation of the model are described and a simple example is presented showing how the model can achieve parsing of natural language. Notwithstanding the apparent paradox of 'decompression by compression', the I…

'Computing' as Information Compression by Multiple Alignment, Unification and Search

This paper argues that the operations of a 'Universal Turing Machine' (UTM) and equivalent mechanisms such as the 'Post Canonical System' (PCS) - which are widely accepted as definitions of the concept of `computing' - may be interpreted as *information compression by multiple alignment, unification and search* (ICMAUS). The motivation for this interpretation is that it suggests ways in which the UTM/PCS model may be augmented in a proposed new computing system designed to exploit the ICMAUS principles as fully as possible. The provision of a relatively sophisticated search mechanism in the pr…

Mathematics and Logic as Information Compression by Multiple Alignment, Unification and Search

This article introduces the conjecture that "mathematics, logic and related disciplines may usefully be understood as information compression (IC) by 'multiple alignment', 'unification' and 'search' (ICMAUS)". As a preparation for the two main sections of the article, concepts of information and information compression are reviewed. Related areas of research are also described including IC in brains and nervous systems, and IC in relation to inductive inference, Minimum Length Encoding and probabilistic reasoning. The ICMAUS concepts and a computer model in which they are embodied are briefly …

Unsupervised Meta Learning for One Shot Title Compression in Voice Commerce

Product title compression for voice and mobile commerce is a well studied problem with several supervised models proposed so far. However these models have 2 major limitations; they are not designed to generate compressions dynamically based on cues at inference time, and they do not transfer well to different categories at test time. To address these shortcomings we model title compression as a meta learning problem where we ask can we learn a title compression model given only 1 example compression? We adopt an unsupervised approach to meta training by proposing an automatic task generation …

从参数到行为：无监督的策略空间压缩

From Parameters to Behaviors: Unsupervised Compression of the Policy Space

本文提出了一种无监督方法，将策略参数空间压缩到低维潜在空间，通过行为重建损失优化生成模型，从而提升连续控制任务中的样本效率和策略表达能力。

无监督行为压缩：通过状态占用匹配学习低维策略流形

Unsupervised Behavioral Compression: Learning Low-Dimensional Policy Manifolds through State-Occupancy Matching

本文提出Occupancy-based Policy Compression方法，通过改进策略压缩框架，以长周期状态空间覆盖替代即时动作匹配，提升行为表示的鲁棒性和泛化能力。

Information Compression, Intelligence, Computing, and Mathematics

This paper presents evidence for the idea that much of artificial intelligence, human perception and cognition, mainstream computing, and mathematics, may be understood as compression of information via the matching and unification of patterns. This is the basis for the "SP theory of intelligence", outlined in the paper and fully described elsewhere. Relevant evidence may be seen: in empirical support for the SP theory; in some advantages of information compression (IC) in terms of biology and engineering; in our use of shorthands and ordinary words in language; in how we merge successive view…

Computing as compression: the SP theory of intelligence

This paper provides an overview of the SP theory of intelligence and its central idea that artificial intelligence, mainstream computing, and much of human perception and cognition, may be understood as information compression. The background and origins of the SP theory are described, and the main elements of the theory, including the key concept of multiple alignment, borrowed from bioinformatics but with important differences. Associated with the SP theory is the idea that redundancy in information may be understood as repetition of patterns, that compression of information may be achieved …

Smart machines and the SP theory of intelligence

These notes describe how the "SP theory of intelligence", and its embodiment in the "SP machine", may help to realise cognitive computing, as described in the book "Smart Machines". In the SP system, information compression and a concept of "multiple alignment" are centre stage. The system is designed to integrate such things as unsupervised learning, pattern recognition, probabilistic reasoning, and more. It may help to overcome the problem of variety in big data, it may serve in pattern recognition and in the unsupervised learning of structure in data, and it may facilitate the management an…

MirrorAlign: A Super Lightweight Unsupervised Word Alignment Model via Cross-Lingual Contrastive Learning

Word alignment is essential for the downstream cross-lingual language understanding and generation tasks. Recently, the performance of the neural word alignment models has exceeded that of statistical models. However, they heavily rely on sophisticated translation models. In this study, we propose a super lightweight unsupervised word alignment model named MirrorAlign, in which bidirectional symmetric attention trained with a contrastive learning objective is introduced, and an agreement loss is employed to bind the attention maps, such that the alignments follow mirror-like symmetry hypothesi…

Knowledge Distillation Methods for Efficient Unsupervised Adaptation Across Multiple Domains

Beyond the complexity of CNNs that require training on large annotated datasets, the domain shift between design and operational data has limited the adoption of CNNs in many real-world applications. For instance, in person re-identification, videos are captured over a distributed set of cameras with non-overlapping viewpoints. The shift between the source (e.g. lab setting) and target (e.g. cameras) domains may lead to a significant decline in recognition accuracy. Additionally, state-of-the-art CNNs may not be suitable for such real-time applications given their computational requirements. A…