As blockchain technology and cryptocurrency become increasingly mainstream, ever-increasing energy costs required to maintain the computational power running these decentralized platforms create a market for more energy-efficient hardware. Photonic cryptographic hash functions, which use photonic integrated circuits to accelerate computation, promise energy efficiency for verifying transactions and mining in a cryptonetwork. Like many analog computing approaches, however, current proposals for photonic cryptographic hash functions that promise similar security guarantees as Bitcoin are susceptible to systematic error, so multiple devices may not reach a consensus on computation despite high numerical precision (associated with low photodetector noise). In this paper, we theoretically and experimentally demonstrate that a more general family of robust discrete analog cryptographic hash functions, which we introduce as LightHash, leverages integer matrix-vector operations on photonic mesh networks of interferometers. The difficulty of LightHash can be adjusted to be sufficiently tolerant to systematic error (calibration error, loss error, coupling error, and phase error) and preserve inherent security guarantees present in the Bitcoin protocol. Finally, going beyond our proof-of-concept, we define a ``photonic advantage'' criterion and justify how recent developments in CMOS optoelectronics (including analog-digital conversion) provably achieve such advantage for robust and digitally-verifiable photonic computing and ultimately generate a new market for decentralized photonic technology.
We propose LightHash , the first feasible photonic cryptographic hash function for blockchain technology using programmable photonic networks. We experimentally evaluate LightHash and assess whether photonic circuits can outperform digital competitors in latency and energy efficiency.
We demonstrate a novel application of online transfer learning for a digital assets trading agent. This agent uses a powerful feature space representation in the form of an echo state network, the output of which is made available to a direct, recurrent reinforcement learning agent. The agent learns to trade the XBTUSD (Bitcoin versus US Dollars) perpetual swap derivatives contract on BitMEX on an intraday basis. By learning from the multiple sources of impact on the quadratic risk-adjusted utility that it seeks to maximise, the agent avoids excessive over-trading, captures a funding profit, and can predict the market's direction. Overall, our crypto agent realises a total return of 350\%, net of transaction costs, over roughly five years, 71\% of which is down to funding profit. The annualised information ratio that it achieves is 1.46.
Due to the rise in popularity of Bitcoin as both a store of wealth and speculative investment, there is an ever-growing demand for automated trading tools to gain an advantage over the market. Although traditionally trading was done by professionals, nowadays a majority of market participants are market-data processing bots due to their inherent advantages in processing large amounts of data, lack of emotions of fear or greed, and predicting market prices through artificial intelligence. A large number of approaches have been brought forward to tackle this task, many of which rely on specially engineered deep learning methods with a focus on specific market conditions. The general limitation of these approaches, however, is the reliance on customized gradient-based methods which limit the scope of possible solutions and don't necessarily generalize well when solving similar problems. This paper proposes a method which uses neuroevolutionary techniques capable of automatically customizing offspring neural networks, generating entire populations of solutions and more thoroughly exploring and parallelizing potential solutions. Our approach uses evolutionary algorithms to evolve increasingly improved populations of neural networks which, based on sentimental and technical analysis data, efficiently predict future market price movements. The effectiveness of this approach is validated by testing the system on both live and historical trading scenarios, and its robustness is tested on other cryptocurrency and stock markets. Experimental results during a 30-day live-trading period show that this method outperformed the buy and hold strategy by over 260%, even while factoring in standard trading fees.
This study will investigate the predictability of a Chaotic time-series data using Reservoir computing (Echo State Network), Deep-Learning(LSTM) and Machine- Learning(Linear, Bayesian, ElasticNetCV , Random Forest, XGBoost Regression and a machine learning Neural Network) on the short (1-day out prediction), medium (5-day out prediction) and long-term (30-day out prediction) pricing of Bitcoin and Ethereum Using a range of machine learning tools, to perform feature selection by permutation importance to select technical indicators on the individual cryptocurrencies, to ensure the datasets are the best for predictions per cryptocurrency while reducing noise within the models. The predictability of these two chaotic time-series is then compared to evaluate the models to find the best fit model. The models are fine-tuned, with hyperparameters, design of the network within the LSTM and the reservoir size within the Echo State Network being adjusted to improve accuracy and speed. This research highlights the effect of the trends within the cryptocurrency and its effect on predictive models, these models will then be optimized with hyperparameter tuning, and be evaluated to compare the models across the two currencies. It is found that the datasets for each cryptocurrency are different, due to the different permutation importance, which does not affect the overall predictability of the models with the short and medium-term predictions having the same models being the top performers. This research confirms that the chaotic data although can have positive results for shortand medium-term prediction, for long-term prediction, technical analysis basedprediction is not sufficient.
А. Е. Мисник, S. K. Krutalevich, Siarhei А. Prakapenka, Peter Borovykh · 5 authors
Neural network is the universal approximator, but its precision highly depends on sufficient set of inputs. Cryptocurrencies have great volatility, due to absence of fundamentals to back up their price. In this paper we analyze approaches to obtain additional parameters for neural networks and explore their impact on its prediction accuracy. This study indicates significant improvement of neural network predictions due to inclusion of wider selection of relevant data points.
The inevitable trend to develop 5G and beyond puts higher requirements on the intelligence of optical networks. In the operation of an optical network, many processes can take advantage of emerging technologies for effective optimization, such as resource allocation, fault operation, traffic classification and prediction and also security convincement.Artificial intelligence (AI) is a method which establishes digital logical topologies following the structure of biological neural network to reproduce its analysis and learning process. Besides, blockchain is also a secure, decentralized architecture that supports the trustworthiness of distributed communications and storage. These two technologies are considered to have great potential in optimizing the intelligent optical network in different aspects.In this paper, we will demonstrate some possible approaches of how AI and blockchain can separately be integrated with optical network to improve its performance.
Neural Networks and Reservoir Computing
Optical Network Technologies
Spectroscopy Techniques in Biomedical and Chemical Research
In this chapter we discuss pseudorandom generators. Loosely speaking, these are efficient deterministic programs that expand short, randomly selected seeds into much longer “pseudorandom” bit sequences (see illustration in Figure 3.1). Pseudorandom sequences are defined as computationally indistinguishable from truly random sequences by efficient algorithms. Hence the notion of computational indistinguishability (i.e., indistinguishability by efficient procedures) plays a pivotal role in our discussion. Furthermore, the notion of computational indistinguishability plays a key role also in subsequent chapters, in particular in the discussions of secure encryption, zero-knowledge proofs, and cryptographic protocols. The theory of pseudorandomness is also applied to functions, resulting in the notion of pseudorandom functions, which is a useful tool for many cryptographic applications. In addition to definitions of pseudorandom distributions, pseudorandom generators, and pseudorandom functions, this chapter contains constructions of pseudorandom generators (and pseudorandom functions) based on various types of one-way functions. In particular, very simple and efficient pseudorandom generators are constructed based on the existence of one-way permutations. We highlight the hybrid technique , which plays a central role in many of the proofs. (For the first use and further discussion of this technique, see Section 3.2.3.) Organization . Basic discussions, definitions, and constructions of pseudorandom generators appear in Sections 3.1–3.4: We start with a motivating discussion (Section 3.1), proceed with a general definition of computational indistinguishability (Section 3.2) next present and discuss definitions of pseudorandom generators (Section 3.3), and finally present some simple constructions (Section 3.4). More general constructions are discussed in Section 3.5.