We demonstrate how word and text embeddings as n-dimensional real vectors used in classical Natural Language Processing (NLP)/AI applications can be mapped to quantum states or quantum embedding representations in Quantum Applications using NLP and AI. Models like fastText [1], [2], GloVe [3], Numberbatch [4], or BERT [5] are common NLP embedding vectors or models that encode semantic properties of words or text fragments. We use these models to evaluate mapping properties and compression rate, as well as information preservation in quantum embeddings. For mapping embedding vectors to quantum states, we use encoding strategies, such as Amplitude Encoding. The encoding strategies allow us to map large dense vectors from the embedding models to compact quantum states using different compression ratios. The compression with Amplitude Encoding can be$2^{n}$to$N$, resulting in a mapping of a 1,024-dimensional vector of reals in the classical environment to a 10-qubit state. The goal of this work is to evaluate these strategies with respect to their compression ratio and measure the preservation of semantic information using similarity scores. We show that the resulting quantum embeddings based on mapped classical computing embeddings exhibit the same relational properties and that there is no significant loss of semantic information in the conversion from classical n-dimensional real vector embeddings to qubit states. We conclude that the experimental results allow us to quantum compute semantic similarities of words or text, reusing existing and freely available embedding models from classical NLP/AI computing.
We demonstrate how word and text embeddings as n-dimensional real vectors used in classical Natural Language Processing (NLP)/AI applications can be mapped to quantum states or quantum embedding representations in Quantum Applications using NLP and AI. Models like fastText [1], [2], GloVe [3], Numberbatch [4], or BERT [5] are common NLP embedding vectors or models that encode semantic properties of words or text fragments. We use these models to evaluate mapping properties and compression rate, as well as information preservation in quantum embeddings. For mapping embedding vectors to quantum states, we use encoding strategies, such as Amplitude Encoding. The encoding strategies allow us to map large dense vectors from the embedding models to compact quantum states using different compression ratios. The compression with Amplitude Encoding can be2ntoN, resulting in a mapping of a 1,024-dimensional vector of reals in the classical environment to a 10-qubit state. The goal of this work is to evaluate these strategies with respect to their compression ratio and measure the preservation of semantic information using similarity scores. We show that the resulting quantum embeddings based on mapped classical computing embeddings exhibit the same relational properties and that there is no significant loss of semantic information in the conversion from classical n-dimensional real vector embeddings to qubit states. We conclude that the experimental results allow us to quantum compute semantic similarities of words or text, reusing existing and freely available embedding models from classical NLP/AI computing.
Word embeddings and vector representations for linguistic units are important in Natural Language Processing (NLP) and AI systems. Engineering of such models involves significant effort, large amounts of data, and costly computations. Many models and systems have been released to the general public, enabling significant research and development in the classical NLP domain. We demonstrate how word and text embeddings as n-dimensional dense vectors of n real numbers in classical computers can be mapped to highly compressed quantum states or quantum embedding using log(n) qubits. In similarity experiments performed on classical and quantum computers, we show no significant information loss that would affect vector distance scores representing words’ semantic similarity. The results show that existing NLP embedding models in classical computing environments can be used in quantum computing. We discuss the issues and limitations of this approach in the context of current quantum computing environments.
We present experimental results from three approaches to encoding classical text, word, or token embeddings, i.e., dense vector representations, as quantum states from state-of-the-art large language models. The experiments focus on identifying the best encoding strategies that achieve the highest compression of classical dense vectors in the number of required qubits, while maintaining the semantic properties of the embeddings trained and reflected in classical AI models. We identify the most optimal interface and data conversion approaches in hybrid classical and quantum computing systems. We demonstrate that the conversion of dense embedding vectors to Hamiltonians is a feasible strategy for a hybrid classicalquantum computing environment and storage of quantum states on classical computers. We report the experimental results of reducing the full embedding-based Hamiltonians using Principal Component Analysis (PCA).
It is important to learn directly from original texts in natural language processing (NLP). Many deep learning (DP) models needing a large number of manually annotated data are not effective in deriving much information from corpora with few annotated labels. Existing methods using unlabeled language information to provide valuable messages consume considerable time and cost. Our provided sentence representation based on quantum computation (called Model I) needs no prior knowledge except word2vec. To reduce some semantic noise caused by the tensor product on the entangled words vector, two improved models (called Model II and Model III) are proposed to reduce the dimensions of the sentence embedding stimulated by Model I. The provided models are evaluated in the STS tasks of 2012, 2014, 2015 and 2016, for a total of 21 corpora. Experimental results show that using quantum entanglement and dimensionality reduction in sentence embedding yields state-of-the-art performances on semantic relations and syntactic structures. Compared to the Pearson correlation coefficient (Pcc) and mean squared error (MSE), the results of 16 out of 16 corpora are better than the results of the comparative methods.
We present a quantum computing approach to analyzing Large Language Model (LLM) embeddings, leveraging complex-valued representations and modeling semantic relationships using quantum mechanical principles. By establishing a direct mapping between LLM semantic spaces and quantum circuits, we demonstrate the feasibility of estimating semantic similarity using quantum hardware. One of the key results is the experimental calculation of cosine similarity between Google Sentence Transformer embeddings using a real quantum computer, providing a tangible demonstration of a quantum approach to semantic analysis. This work reveals a connection between LLMs and quantum mechanics, suggesting that these principles can offer new perspectives on semantic representation and processing, and paving the way for future development of quantum algorithms for natural language processing.
Large language and sentence-embedding models provide rich semantic representations, but their high dimensionality poses a challenge for near-term quantum machine learning (QML), where quantum circuits can process only a limited number of input features. We investigate a hybrid quantum-classical pipeline that transforms high-dimensional sentence embeddings into compact representations for variational quantum classification. The workflow combines a pretrained sentence-embedding model, dimensionality reduction, angle encoding, a variational quantum circuit (VQC), and a classical decision layer. We systematically compare principal component analysis (PCA), neighborhood components analysis (NCA), and linear discriminant analysis (LDA), covering both unsupervised and supervised dimensionality reduction. Using the TREC question-classification dataset, we study the relationship between representation dimensionality, information retention, qubit count, and classification performance. Preliminary PCA experiments reveal a strong information bottleneck: reducing 768-dimensional embeddings to 3, 4, 5, and 8 dimensions retains about 8.2%, 10.2%, 11.9%, and 16.4% of the variance, with corresponding classification accuracies of 50.3%, 51.2%, 57.9%, and 63.4%. In contrast, supervised reduction is substantially more efficient. LDA reaches 85.3% accuracy and NCA reaches 83.1% using only 5 dimensions, under a leakage-free cross-validation protocol, compared with 85.1% for a full 384-dimensional classical baseline. These results indicate that supervised dimensionality reduction can preserve task-relevant information far more effectively than variance-based compression, making compact representations a promising route toward practical hybrid quantum-classical NLP models.
Over the last decade, representation learning, which embeds complex information extracted from large amounts of data into dense vector spaces, has emerged as a key technique in machine learning. Among other applications, it has been a key building block for large language models and advanced computer vision systems based on contrastive learning. A core component of representation learning systems is the projection head, which maps the original embeddings into different, often compressed spaces, while preserving the similarity relationship between vectors. In this paper, we propose a quantum-inspired projection head that includes a corresponding quantum-inspired similarity metric. Specifically, we map classical embeddings onto quantum states in Hilbert space and introduce a quantum circuit-based projection head to reduce embedding dimensionality. To evaluate the effectiveness of this approach, we extended the BERT language model by integrating our projection head for embedding compression. We compared the performance of embeddings, which were compressed using our quantum-inspired projection head, with those compressed using a classical projection head on information retrieval tasks using the TREC 2019 and TREC 2020 Deep Learning benchmarks. The results demonstrate that our quantum-inspired method achieves competitive performance relative to the classical method while utilizing 32 times fewer parameters. Furthermore, when trained from scratch, it notably excels, particularly on smaller datasets. This work not only highlights the effectiveness of the quantum-inspired approach but also emphasizes the utility of efficient, ad hoc low-entanglement circuit simulations within neural networks as a powerful quantum-inspired technique.