Ремедиум. 2019; : 36-43
Современный передовой уровень искусственного интеллекта для умной медицины
https://doi.org/10.21518/1561-5936-2019-04-36-43Аннотация
На данном этапе искусственный интеллект уже не является только обсуждаемой темой. Это вполне реальные технологии, основанные преимущественно на искусственных нейронных сетях. Для их обучения используется принцип Павлова, сформулированный В. Л. Дуниным-Барковским. Математики павловское учение с подкреплением называют Deep Reinforcement Learning. ИИ разделяют на компьютерное зрение (Computer Vision), т е. распознавание и генерацию изображений; распознавание и синтез речи (Speech Recognition and Synthesis); обработку естественного языка (Natural Language Processing, NLP); графовый логический ИИ, миварную логическую технологию. Все это по отдельности - узконаправленный ИИ. А общий интеллект, равный человеку, пока не создан. Такой ИИ должен включать в себя все технологии. С учетом социальной и лингвистической природы появления интеллекта разработчики очень много внимания уделяют отшлифовке алгоритмов NLP и мультиагентной среды. К сожалению, параллельно с прогрессом в развитии нейросетей возникло такое явление, как состязательные атаки, которые, используя тот же механизм обучения, заставляют натренированную нейросеть делать ошибки. Этот факт подвергает сомнению будущее нейросетей в повседневной медицине. Среда для ИИ - это большие данные и датасеты. Европейские эксперты уже озадачились регулированием больших данных с точки зрения безопасного развития как медицины, так и фармацевтической сферы. Несмотря на сложности и отсутствие четких правил, ИИ активно внедряется в частный сектор медицины, создав уже три новые бизнес-модели.
Список литературы
1. Bartol T.M., Bromer C., Kinney J., Chirillo M.A., Bourne J.N., Sejnowski T.J. et al. Nanoconnectomic upper bound on the variability of synaptic plasticity. eLife J. 2015;4:e10778. DOI: 10.7554/eLife.10778.
2. Howard D., Eiben A.E., Kennedy D.F., Mouret J.-B., Valencia P., Winkler D. Evolving embodied intelligence from materials to machines. Nature Machine Intelligence. 2019;1:12-19. DOI. org/10.1038/s42256-018-0009-9.
3. Дунин-Барковский В.Л., Соловьева К.П. Принцип Павлова в проблеме обратного конструирования мозга. XVIII Международная конференция «Нейроинформатика-2016». Сборник научных трудов, ч. 1. М.: Национальный исследовательский ядерный университет «МИФИ», 2016:11-23.
4. Dunin-Barkowski W., Solovyeva K. Pavlov Principle and Brain Reverse Engineering. IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology, CIBCB- 2018. Saint Louis, Missouri, USA. 2018; Paper #37: 1-5. DOI: 10.1109/CIBCB.2018.8404975.
5. Shakirov V.V., Solovyeva K.P., Dunin-Barkowski W.L. Review of State-of-the-Art in Deep Learning Artificial Intelligence. Optical Memory and Neural Networks. 2018;27(2):65-80. DOI: 10.3103/S1060992X18020066.
6. Dunin-Barkowski W.L., Shakirov V.V. A Way toward Human Level Artificial Intelligence. Optical Memory and Neural Networks. 2019;28(1):21-26. DOI: 10.3103/S1060992X19010041.
7. Varlamov O.O. Wi!Mi Expert System Shell as the Novel Tool for Building Knowledge-Based Systems with Linear Computational Complexity. The International Review of Automatic Control (IREACO). 2018;11(6):314-325. DOI.org/10.15866/ireaco.v11i6.15855.
8. Varlamov O.O., Chuvikov D.A., Adamova L.E., Kolesnichenko O.Yu., Petrov M.A., Zabolotskaya I.K., Zhilina T.N. Logical, Philosophical and Ethical Aspects of AI in Medicine. International Conference on Computer Science and Information Technology (ICCSIT-2018), International Journal of Machine Learning and Computing. 2019. В печати.
9. Booth Jo., Booth Ja. Marathon Environments: Multi-Agent Continuous Control Benchmarks in a Modern Video Game Engine. arXiv:1902.09097v1 [cs.AI] 25 Feb 2019.
10. Pathak D., Lu C., Darrell T., Isola P., Efros A.A. Learning to Control Self-Assembling Morphologies: A Study of Generalization via Modularity. arXiv:1902.05546v1 [cs.LG] 14 Feb 2019.
11. Tassa Y., Doron Y., Muldal A., Erez T., Li Y., Lillicrap T. et al. Deepmind control suite. arXiv:1801.00690v1 [cs.AI] 2 Jan 2018.
12. Wang R., Lehman J., Clune J., Stanley K.O. Paired Open-Ended Trailblazer (POET): Endlessly Generating Increasingly Complex and Diverse Learning Environments and Their Solutions. arXiv:1901.01753v3 [cs.NE] 21 Feb 2019.
13. Gopalakrishnan A., Mali A., Kifer D., Lee Giles C., Ororbia A.G. A Neural Temporal Model for Human Motion Prediction. arXiv:1809.03036v4 [cs.CV] 6 Dec 2018.
14. Hernandez-Ruiz A., Gall J., Moreno-Noguer F. Human Motion Prediction via Spatio-Temporal Inpainting. arXiv:1812.05478v1 [cs.CV] 13 Dec 2018.
15. Qiu J., Huang G., Lee T.S. A Neurally-Inspired Hierarchical Prediction Network for Spatiotemporal Sequence Learning and Prediction. arXiv:1901.09002v1 [cs.NE] 25 Jan 2019.
16. Radford A., Jozefowicz R., Sutskever I. Learning to Generate Reviews and Discovering Sentiment. arXiv:1704.01444v2 [cs.LG] 6 Apr 2017.
17. Radford A., Narasimhan K., Salimans T., Sutskever I. Improving Language Understanding by Generative Pre-Training, 2018. URL: https://s3-us-west-2.amazonaws.com/openai-assets/research-covers/language-unsupervised/language_understanding_paper.pdf.
18. Radford A., Wu J., Child R., Luan D., Amodei D., Sutskever I. Language Models are Unsupervised Multitask Learners, 2019. https://github.com/openai/gpt-2.
19. Yogatama D., De Masson d’Autume C., Connor J., Kocisky T., Chrzanowski M., Kong L. et al. Learning and Evaluating General Linguistic Intelligence. arXiv:1901.11373v1 [cs.LG] 31 Jan 2019.
20. Aharoni R., Johnson M., Firat O. Massively Multilingual Neural Machine Translation. arXiv:1903.00089v1 [cs.CL] 28 Feb 2019.
21. Lample G., Conneau A. Cross-lingual Language Model Pretraining. arXiv:1901.07291v1 [cs.CL] 22 Jan 2019.
22. Nachmani E., Wolf L. Unsupervised Polyglot Text To Speech. arXiv:1902.02263v1 [cs.LG] 6 Feb 2019.
23. Haque A, Guo M, Verma P, Fei-Fei L. Audio-Linguistic Embeddings for Spoken Sentences. arXiv:1902.07817v1 [cs.SD] 20 Feb 2019.
24. Gupta A., Vedaldi A., Zisserman A. Learning to Read by Spelling: Towards Unsupervised Text Recognition. arXiv:1809.08675v2 [cs.CV] 9 Dec 2018.
25. Finlayson S.G., Bowers J.D., Ito J., Zittrain J.L., Beam A.L., Kohane I.S. Adversarial attacks on medical machine learning. Science, 2019;363(6433):1287-1289. DOI: 10.1126/science.aaw4399.
26. Finlayson S.G., Chung H.W., Kohane I.S., Beam A.L. Adversarial Attacks Against Medical Deep Learning Systems. arXiv:1804.05296v3 [cs.CR] 4 Feb 2019.
27. Kolesnichenko Yu., Kolesnichenko O., Smorodin G. 3-Dimensional Vector Analysis of 2-Dimensional Ultrasound Diagnostic Images. 21st Conference of Open Innovations Association FRUCT, University of Helsinki, Finland, 2017:428-434.
28. HMA-EMA Joint Big Data Taskforce, Summary report. Heads of Medicines Agencies EU, European Medicines Agency. EMA/105321/2019.13 February 2019, 48.
Remedium. 2019; : 36-43
MODERN ADVANCED ARTIFICIAL INTELLIGENCE FOR SMART MEDICINE
https://doi.org/10.21518/1561-5936-2019-04-36-43Abstract
Artificial Intelligence is no longer just the topic of discussion. Today this technology is mostly based on Artificial Neural Networks. Pavlov Principle formulated by W.L. Dunin-Barkowski is used for their training. Mathematics compared Pavlov's doctrine with Deep Reinforcement Learning. AI technologies are divided into Computer Vision, images recognition and generation; Speech Recognition and Synthesis; Natural Language Processing; Graph Logic AI, MIVAR logic technology. All of this separately is Narrow AI. Artificial General Intelligence, equal to human, hasn’t been created yet. AGI should include all mentioned technologies. Given social and linguistic nature of the intelligence emergence, developers are paying attention to NLP algorithms and multi-agent environment. Simultaneously with the development of neural networks, adversary attacks emerged, which using the same learning mechanism force a trained neural network to make mistakes. This fact calls in question the future of neural networks in medicine. Big Data and data sets are the environment for AI. European experts have already begun to regulate Big Data for safe Health Care and drugs creation. Despite the difficulties and lack of clear rules, AI is actively being introduced into the private medicine. Due to AI the three new business models have already been created.
References
1. Bartol T.M., Bromer C., Kinney J., Chirillo M.A., Bourne J.N., Sejnowski T.J. et al. Nanoconnectomic upper bound on the variability of synaptic plasticity. eLife J. 2015;4:e10778. DOI: 10.7554/eLife.10778.
2. Howard D., Eiben A.E., Kennedy D.F., Mouret J.-B., Valencia P., Winkler D. Evolving embodied intelligence from materials to machines. Nature Machine Intelligence. 2019;1:12-19. DOI. org/10.1038/s42256-018-0009-9.
3. Dunin-Barkovskii V.L., Solov'eva K.P. Printsip Pavlova v probleme obratnogo konstruirovaniya mozga. XVIII Mezhdunarodnaya konferentsiya «Neiroinformatika-2016». Sbornik nauchnykh trudov, ch. 1. M.: Natsional'nyi issledovatel'skii yadernyi universitet «MIFI», 2016:11-23.
4. Dunin-Barkowski W., Solovyeva K. Pavlov Principle and Brain Reverse Engineering. IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology, CIBCB- 2018. Saint Louis, Missouri, USA. 2018; Paper #37: 1-5. DOI: 10.1109/CIBCB.2018.8404975.
5. Shakirov V.V., Solovyeva K.P., Dunin-Barkowski W.L. Review of State-of-the-Art in Deep Learning Artificial Intelligence. Optical Memory and Neural Networks. 2018;27(2):65-80. DOI: 10.3103/S1060992X18020066.
6. Dunin-Barkowski W.L., Shakirov V.V. A Way toward Human Level Artificial Intelligence. Optical Memory and Neural Networks. 2019;28(1):21-26. DOI: 10.3103/S1060992X19010041.
7. Varlamov O.O. Wi!Mi Expert System Shell as the Novel Tool for Building Knowledge-Based Systems with Linear Computational Complexity. The International Review of Automatic Control (IREACO). 2018;11(6):314-325. DOI.org/10.15866/ireaco.v11i6.15855.
8. Varlamov O.O., Chuvikov D.A., Adamova L.E., Kolesnichenko O.Yu., Petrov M.A., Zabolotskaya I.K., Zhilina T.N. Logical, Philosophical and Ethical Aspects of AI in Medicine. International Conference on Computer Science and Information Technology (ICCSIT-2018), International Journal of Machine Learning and Computing. 2019. V pechati.
9. Booth Jo., Booth Ja. Marathon Environments: Multi-Agent Continuous Control Benchmarks in a Modern Video Game Engine. arXiv:1902.09097v1 [cs.AI] 25 Feb 2019.
10. Pathak D., Lu C., Darrell T., Isola P., Efros A.A. Learning to Control Self-Assembling Morphologies: A Study of Generalization via Modularity. arXiv:1902.05546v1 [cs.LG] 14 Feb 2019.
11. Tassa Y., Doron Y., Muldal A., Erez T., Li Y., Lillicrap T. et al. Deepmind control suite. arXiv:1801.00690v1 [cs.AI] 2 Jan 2018.
12. Wang R., Lehman J., Clune J., Stanley K.O. Paired Open-Ended Trailblazer (POET): Endlessly Generating Increasingly Complex and Diverse Learning Environments and Their Solutions. arXiv:1901.01753v3 [cs.NE] 21 Feb 2019.
13. Gopalakrishnan A., Mali A., Kifer D., Lee Giles C., Ororbia A.G. A Neural Temporal Model for Human Motion Prediction. arXiv:1809.03036v4 [cs.CV] 6 Dec 2018.
14. Hernandez-Ruiz A., Gall J., Moreno-Noguer F. Human Motion Prediction via Spatio-Temporal Inpainting. arXiv:1812.05478v1 [cs.CV] 13 Dec 2018.
15. Qiu J., Huang G., Lee T.S. A Neurally-Inspired Hierarchical Prediction Network for Spatiotemporal Sequence Learning and Prediction. arXiv:1901.09002v1 [cs.NE] 25 Jan 2019.
16. Radford A., Jozefowicz R., Sutskever I. Learning to Generate Reviews and Discovering Sentiment. arXiv:1704.01444v2 [cs.LG] 6 Apr 2017.
17. Radford A., Narasimhan K., Salimans T., Sutskever I. Improving Language Understanding by Generative Pre-Training, 2018. URL: https://s3-us-west-2.amazonaws.com/openai-assets/research-covers/language-unsupervised/language_understanding_paper.pdf.
18. Radford A., Wu J., Child R., Luan D., Amodei D., Sutskever I. Language Models are Unsupervised Multitask Learners, 2019. https://github.com/openai/gpt-2.
19. Yogatama D., De Masson d’Autume C., Connor J., Kocisky T., Chrzanowski M., Kong L. et al. Learning and Evaluating General Linguistic Intelligence. arXiv:1901.11373v1 [cs.LG] 31 Jan 2019.
20. Aharoni R., Johnson M., Firat O. Massively Multilingual Neural Machine Translation. arXiv:1903.00089v1 [cs.CL] 28 Feb 2019.
21. Lample G., Conneau A. Cross-lingual Language Model Pretraining. arXiv:1901.07291v1 [cs.CL] 22 Jan 2019.
22. Nachmani E., Wolf L. Unsupervised Polyglot Text To Speech. arXiv:1902.02263v1 [cs.LG] 6 Feb 2019.
23. Haque A, Guo M, Verma P, Fei-Fei L. Audio-Linguistic Embeddings for Spoken Sentences. arXiv:1902.07817v1 [cs.SD] 20 Feb 2019.
24. Gupta A., Vedaldi A., Zisserman A. Learning to Read by Spelling: Towards Unsupervised Text Recognition. arXiv:1809.08675v2 [cs.CV] 9 Dec 2018.
25. Finlayson S.G., Bowers J.D., Ito J., Zittrain J.L., Beam A.L., Kohane I.S. Adversarial attacks on medical machine learning. Science, 2019;363(6433):1287-1289. DOI: 10.1126/science.aaw4399.
26. Finlayson S.G., Chung H.W., Kohane I.S., Beam A.L. Adversarial Attacks Against Medical Deep Learning Systems. arXiv:1804.05296v3 [cs.CR] 4 Feb 2019.
27. Kolesnichenko Yu., Kolesnichenko O., Smorodin G. 3-Dimensional Vector Analysis of 2-Dimensional Ultrasound Diagnostic Images. 21st Conference of Open Innovations Association FRUCT, University of Helsinki, Finland, 2017:428-434.
28. HMA-EMA Joint Big Data Taskforce, Summary report. Heads of Medicines Agencies EU, European Medicines Agency. EMA/105321/2019.13 February 2019, 48.
