arXiv subjects

Rushikesh Ubale

Publications and source records attributed to Rushikesh Ubale.

3 recordsLinked to original sources

arXiv2025

Quantum Similarity-Driven QUBO Framework for Multi-Period Supply Chain Allocation using Time-Multiplexed Coherent Ising Machines and Simulated Quantum Annealing

Multi-period stock-keeping unit (SKU) allocation in supply chains is a combinatorial optimization problem that is both NP-hard and operationally critical, requiring simultaneous attention to profitability, feasibility, and diversity. Quadratic unconstrained binary optimization (QUBO) provides a principled framework for such tasks, yet prior studies often rely on simplified assumptions or omit real operational constraints. This work proposes a hybrid QUBO framework integrating three advances: (i) a quantum-derived similarity kernel, obtained from a variational RX embedding, to discourage redundant SKU selections; (ii) exact per-period capacity enforcement via slack-bit encoding to maintain feasibility; and (iii) execution on a time-multiplexed Coherent Ising Machine (CIM) benchmarked against simulated quantum annealing (SQA) and classical optimization algorithms. The resulting model, with over one million quadratic terms and about 4,100 variables, captures profit, risk, and capacity interactions within a unified formulation. On a dataset of 500 SKUs across eight planning periods, Quanfluence's CIM achieved an energy of minus 2.95 times 10 to the power of 16, producing robust solutions with 288 distinct SKUs (approximately 60 percent of the catalog), 226,813 allocated units, and 12.75 million dollars profit, all with zero capacity violations. These results demonstrate that hybrid quantum-classical QUBO methods can deliver feasible and profitable supply-chain allocations at an industrial scale.

quant-ph↗

arXiv2025

Toward Practical Quantum Machine Learning: A Novel Hybrid Quantum LSTM for Fraud Detection

We present a novel hybrid quantum-classical neural network architecture for fraud detection that integrates a classical Long Short-Term Memory (LSTM) network with a variational quantum circuit. By leveraging quantum phenomena such as superposition and entanglement, our model enhances the feature representation of sequential transaction data, capturing complex non-linear patterns that are challenging for purely classical models. A comprehensive data preprocessing pipeline is employed to clean, encode, balance, and normalize a credit card fraud dataset, ensuring a fair comparison with baseline models. Notably, our hybrid approach achieves per-epoch training times in the range of 45-65 seconds, which is significantly faster than similar architectures reported in the literature, where training typically requires several minutes per epoch. Both classical and quantum gradients are jointly optimized via a unified backpropagation procedure employing the parameter-shift rule for the quantum parameters. Experimental evaluations demonstrate competitive improvements in accuracy, precision, recall, and F1 score relative to a conventional LSTM baseline. These results underscore the promise of hybrid quantum-classical techniques in advancing the efficiency and performance of fraud detection systems. Keywords: Hybrid Quantum-Classical Neural Networks, Quantum Computing, Fraud Detection, Hybrid Quantum LSTM, Variational Quantum Circuit, Parameter-Shift Rule, Financial Risk Analysis

quant-ph↗

arXiv2025

End-to-End Portfolio Optimization with Quantum Annealing

Hybrid-quantum classical optimization has emerged as a promising direction for addressing financial decision problems under current quantum hardware constraints. In this work we present a practical end-to-end portfolio optimization pipeline that combines (i) a continuous mean-variance and Sharpe-ratio formulation, (ii) a QUBO/CQM-based discrete asset selection stage solved using D-Wave's hybrid quantum annealing solver, (iii) classical convex optimization for computing optimal asset weights, and (iv) a quarterly rebalancing mechanism. Rather than claiming quantum advantage, our goal is to evaluate the feasibility and integration of these components within a deployable financial workflow. We empirically compare our hybrid pipeline against a fund manager in real time and indexes used in Indian stock market. The results indicate that the proposed framework can construct diversified portfolios and achieve competitive returns. We also report computational considerations and scalability observations drawn from the hybrid solver behaviour. While the experiments are limited to moderate sized portfolios dictated by current annealing hardware and QUBO embedding constraints, the study illustrates how quantum assisted selection and classical allocation can be combined coherently in a real-world setting. This work emphasizes methodological reproducibility and practical applicability, and aims to serve as a step toward larger-scale financial optimization workflows as quantum annealers continue to mature.

quant-ph↗