Dr Yuhua Li

Darllenydd Dysgu Peiriant

: Ar gael fel goruchwyliwr ôl-raddedig

Rwyf wedi cynnal ymchwil sylfaenol a chymhwysol mewn dysgu peiriannau, adnabod patrwm, gwyddor data, dadansoddiad tebygrwydd semantig a monitro cyflyrau. Rwy'n arwain y Data Analytics a'r Grŵp Ymchwil Dysgu Peiriant.

Mae fy mhrofiad o ddysgu peirianyddol a chydnabod patrymau yn cynnwys dulliau ystadegol, geometrig a rhwydweithiau niwral ar gyfer dethol nodwedd / patrwm a dadansoddi data, darganfod gwybodaeth a chasgliad.

Mae fy nghyfraniad i ddysgu peirianyddol yn cynnwys datblygu dulliau canfod anghysondeb/newydd-deb ar gyfer systemau diogelwch / cenhadaeth-gritig, sydd â data/gwybodaeth gyfyngedig neu ddim gwybodaeth am ddigwyddiadau prin, a thechnegau dethol arsylwi llawn gwybodaeth ar gyfer synwyryddion/mesuriadau optimeiddio lleoliad ar gyfer problemau fel monitro effeithiol a rheoli prosesau. Mae fy ngwaith wedi ysgogi ymchwilwyr eraill i ddatblygu algorithmau AI newydd, eu defnyddio fel meincnodau a'u mabwysiadu mewn cynhyrchion.

Rwyf wedi arwain a chynnal prosiectau ymchwil a ariennir gan y llywodraeth, elusen a diwydiant. Rwyf wedi cydweithio ar brosiectau ymchwil gyda chwmnïau cenedlaethol a rhyngwladol o wahanol feintiau. Mae fy ymchwil wedi'i gymhwyso i ddatrys problemau mewn gweithgynhyrchu digidol, monitro cyflyrau, peirianneg ariannol, a phroblemau eraill yn y byd go iawn.

Ymgysylltu Allanol

Aelod o Goleg Adolygu Cymheiriaid EPSRC
Cynghorydd Academaidd Comisiwn Ysgoloriaethau'r Gymanwlad y DU
Aelod o Gyngor Cynghori ar Arloesi Cymru
Golygydd Cyswllt Trafodion IEEE ar Rwydweithiau Nerfol a Systemau Dysgu

Date
Type

2024

Anggoro, A., Corcoran, P., De Widt, D. and Li, Y. 2024. Harmonized system code classification using supervised contrastive learning with sentence BERT and multiple negative ranking loss. Data Technologies and Applications (10.1108/DTA-01-2024-0052)
Lewis-Cheetham, J., Li, Y., Liberatore, F. and Wang, Q. 2024. The impact of transaction costs on forecast-based trading strategy performance. Presented at: CIFEr 2024: IEEE Symposium on Computational Intelligence for Financial Engineering and Economics, Hoboken, New Jersey, USA, 22-23 October 2024Proceedings of the IEEE Symposium on Computational Intelligence for Financial Engineering and Economics. IEEE pp. 1-8., (10.1109/CIFEr62890.2024.10772810)
Bent, G. A. et al. 2024. The transformative potential of vector symbolic architecture for cognitive processing at the network edge. Presented at: Security and Defence 2024, Edinburgh, UK, 16-20 September 2024 Presented at Bouma, H. et al. eds.Proceedings Artificial Intelligence for Security and Defence Applications II, Vol. 13206. SPIE pp. 1320612., (10.1117/12.3030949)
Bent, G. A. et al. 2024. A demonstration of vector symbolic architecture as an effective integrated technology for AI at the network edge. Presented at: Artificial Intelligence for Security and Defence Applications II, Edinburgh, UK, 17–19 September 2024 Presented at Bouma, H. et al. eds.Proceedings of Artificial Intelligence for Security and Defence Applications II. Vol. 132061. SPIE pp. 39-55., (10.1117/12.3030952)
Wang, H., Marshall, A., Jones, D. and Li, Y. 2024. Improving high-frequency details in cerebellum for brain MRI super-resolution. Presented at: Conference on ICT Solutions for eHealth (ICTS4eHealth 2024), Paris, France, 26 - 29 June 20242024 IEEE Symposium on Computers and Communications (ISCC). IEEE pp. 1-7., (10.1109/ISCC61673.2024.10733580)
Zhang, M., Treder, M., Marshall, D. and Li, Y. 2024. Explaining the predictions of kernel SVM models for neuroimaging data analysis. Expert Systems with Applications 251, article number: 123993. (10.1016/j.eswa.2024.123993)
Wu, F., Liu, A., Chen, J. and Li, Y. 2024. Analysing network dynamics: The contagion effects of SVB's collapse on the US tech industry. Journal of Risk and Financial Management 17(10), article number: 427. (10.3390/jrfm17100427)
Zhang, M., Treder, M., Marshall, A. and Li, Y. 2024. Fast explanation of RBF-Kernel SVM models using activation patterns. Presented at: International Joint Conference on Neural Networks, Yokohama, Japan, 30 June – 5 July 2024Proceedings of IJCNN. IEEE pp. 1-8., (10.1109/IJCNN60899.2024.10650697)
Alqurashi, N., Li, Y. and Sidorov, K. 2024. Improving speech emotion recognition through hierarchical classification and text integration for enhanced emotional analysis and contextual understanding. Presented at: International Joint Conference on Neural Networks, Yokohama, Japan, 30 June – 5 July 2024Proceedings of IJCNN. IEEE pp. 1-8., (10.1109/IJCNN60899.2024.10650087)
Alqurashi, N., Li, Y., Sidorov, K. and Marshall, A. 2024. Decision fusion based multimodal hierarchical method for speech emotion recognition from audio and text. Presented at: European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, Bruges, Belgium, 9-11 October 2024.
Li, S., Li, Y. and Perera, C. 2024. Mobile sensing within smart buildings: A survey. Technical Report.

2023

Wang, H., Treder, M., Marshall, D., Jones, D. and Li, Y. 2023. A skewed loss function for correcting predictive bias in brain age prediction. IEEE Transactions on Medical Imaging 42(6), pp. 1577-1589. (10.1109/TMI.2022.3231730)
Zhang, F., Zhao, H., Li, Y., Wu, Y. and Sun, X. 2023. CBA-GAN: Cartoonization style transformation based on the convolutional attention module. Computers and Electrical Engineering 106, article number: 108575. (10.1016/j.compeleceng.2022.108575)
Anggoro, A. W., Corcoran, P., De Widt, D. and Li, Y. 2023. Using DistilBERT to assign HS codes to international trading transactions. Presented at: World Conference on Information Systems and Technologies, Pisa, Italy, 4 - 6 April 2023.

2022

Sidorowicz, T., Peres, P. and Li, Y. 2022. A novel approach for cross-selling insurance products using positive unlabelled learning. Presented at: International Joint Conference on Neural Networks, Padua - Italy, 18-23 July 20222022 International Joint Conference on Neural Networks (IJCNN). IEEE, (10.1109/IJCNN55064.2022.9892762)
Bent, G., Simpkin, C., Li, Y. and Preece, A. 2022. Energy efficient spiking neural network neuromorphic processing to enable decentralised service workflow composition in support of multi-domain operations. Presented at: SPIE Defense + Commercial Sensing 2022, Orlando, Florida, United States, 3 April - 13 June 2022 Presented at Pham, T. and Solomon, L. eds.Artificial Intelligence and Machine Learning for Multi-Domain Operations Applications, Vol. 12113. SPIE pp. 121131M., (10.1117/12.2617362)
Bent, G., Simpkin, C., Li, Y. and Preece, A. 2022. Hyperdimensional computing using time-to-spike neuromorphic circuits. Presented at: 2022 IEEE World Congress on Computational Intelligence (WCCI), Padova, Italy, 18-23 July 2022. IEEE

2021

Sohail, M., Peres, P. and Li, Y. 2021. Feature importance analysis for customer management of insurance products. Presented at: 2021 International Joint Conference on Neural Networks (IJCNN), Virtual, 18-22 July 2021.

2020

Zhao, H., Zheng, J., Wang, Y., Yuan, X. and Li, Y. 2020. Portrait style transfer using deep convolutional neural networks and facial segmentation. Computers and Electrical Engineering 85, article number: 106655. (10.1016/j.compeleceng.2020.106655)
Taherkhani, A., Belatreche, A., Li, Y., Cosma, G., Maguire, L. P. and McGinnity, T. 2020. A review of learning in biologically plausible spiking neural networks. Neural Networks 122, pp. 253-272. (10.1016/j.neunet.2019.09.036)

2018

Taherkhani, A., Belatreche, A., Li, Y. and Maguire, L. P. 2018. A supervised learning algorithm for learning precise timing of multiple spikes in multilayer spiking neural networks. IEEE Transactions on Neural Networks and Learning Systems 29(11), pp. 5394-5407. (10.1109/TNNLS.2018.2797801)
Liu, S. et al. 2018. Quantitative analysis of breast cancer diagnosis using a probabilistic modelling approach. Computers in Biology and Medicine 92, pp. 168-175. (10.1016/j.compbiomed.2017.11.014)

2017

Shynkevich, Y., McGinnity, T., Coleman, S. A., Belatreche, A. and Li, Y. 2017. Forecasting price movements using technical indicators: Investigating the impact of varying input window length. Neurocomputing 264, pp. 71--88. (10.1016/j.neucom.2016.11.095)
Zhai, J., Cao, Y., Yao, Y., Ding, X. and Li, Y. 2017. Coarse and fine identification of collusive clique in financial market. Expert Systems with Applications 69, pp. 225-238. (10.1016/j.eswa.2016.10.051)
Zhai, J., Cao, Y., Yao, Y., Ding, X. and Li, Y. 2017. Computational intelligent hybrid model for detecting disruptive trading activity. Decision Support Systems 93, pp. 26--41. (10.1016/j.dss.2016.09.003)

2016

Cao, Y., Li, Y., Coleman, S., Belatreche, A. and McGinnity, T. M. 2016. Detecting wash trade in financial market using digraphs and dynamic programming. IEEE Transactions on Neural Networks and Learning Systems 27(11), pp. 2351-2363. (10.1109/TNNLS.2015.2480959)
Raza, H., Cecotti, H., Li, Y. and Prasad, G. 2016. Adaptive learning with covariate shift-detection for motor imagery-based brain–computer interface. Soft Computing 20(8), pp. 3085--3096. (10.1007/s00500-015-1937-5)
Liu, J., Harkin, J., Li, Y. and Maguire, L. P. 2016. Fault-tolerant networks-on-chip routing with coarse and fine-grained look-ahead. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems 35(2), pp. 260-273. (10.1109/TCAD.2015.2459050)

2015

Liu, J., Harkin, J., Li, Y. and Maguire, L. 2015. Low cost fault-tolerant routing algorithm for Networks-on-Chip. Microprocessors and Microsystems 39(6), pp. 358-372. (10.1016/j.micpro.2015.06.002)
Raza, H., Prasad, G. and Li, Y. 2015. EWMA model based shift-detection methods for detecting covariate shifts in non-stationary environments. Pattern Recognition 48(3), pp. 659-669. (10.1016/j.patcog.2014.07.028)
Ding, X., Li, Y., Belatreche, A. and Maguire, L. P. 2015. Novelty detection using level set methods. IEEE Transactions on Neural Networks and Learning Systems 26(3), pp. 576-588. (10.1109/TNNLS.2014.2320293)
Cao, Y., Li, Y., Coleman, S., Belatreche, A. and McGinnity, T. M. 2015. Adaptive hidden Markov model with anomaly states for price manipulation detection. IEEE Transactions on Neural Networks and Learning Systems 26(2), pp. 318-330. (10.1109/TNNLS.2014.2315042)
Taherkhani, A., Belatreche, A., Li, Y. and Maguire, L. P. 2015. DL-ReSuMe: A delay learning-based remote supervised method for spiking neurons. IEEE Transactions on Neural Networks and Learning Systems 26(12), pp. 3137-3149. (10.1109/TNNLS.2015.2404938)

2014

Liu, J., Harkin, J., Li, Y., Maguire, L. and Linares-Barranco, A. 2014. Low overhead monitor mechanism for fault-tolerant analysis of NoC. Presented at: 8th International Symposium On Embedded Multicore/manycore Socs, Aizu-Wakamatsu, Japan, 23-25 Sep 20148th International Symposium on Embedded Multicore/manycore Socs. IEEE pp. 189-196., (10.1109/MCSoC.2014.35)
Liu, J., Harkin, J., Li, Y. and Maguire, L. 2014. Online fault detection for Networks-on-Chip interconnect. Presented at: 2014 NASA/ESA Conference on Adaptive Hardware and Systems (AHS), Leicester, UK, 14-17 Jul 20142014 NASA/ESA Conference on Adaptive Hardware and Systems (AHS 2014). Piscataway, NJ: IEEE pp. 31-38., (10.1109/AHS.2014.6880155)
Ding, X., Li, Y., Belatreche, A. and Maguire, L. P. 2014. An experimental evaluation of novelty detection methods. Neurocomputing 135, pp. 313-327. (10.1016/j.neucom.2013.12.002)
Raza, H., Prasad, G. and Li, Y. 2014. Dataset shift detection in non-stationary environments using EWMA charts. Presented at: 2013 IEEE International Conference on Systems, Man, and Cybernetics, Manchester, UK, 13-16 Oct 20132013 IEEE International Conference on Systems, Man, and Cybernetics. Piscataway, New Jersey: IEEE pp. 3151-3156., (10.1109/SMC.2013.537)
Ding, X., Li, Y., Belatreche, A. and Maguire, L. P. 2014. Novelty detection using level set methods with adaptive boundaries. Presented at: SMC 2013, Manchester, UK, 13-16 Oct 20132013 IEEE International Conference on Systems, Man and Cybernetics (SMC 2013). Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics Piscataway, NJ: IEEE pp. 3020-3025., (10.1109/SMC.2013.515)
Cao, Y., Li, Y., Coleman, S., Belatreche, A. and McGinnity, T. M. 2014. A hidden Markov model with abnormal states for detecting stock price manipulation. Presented at: SMC 2013, Manchester, UK, 13-16 Oct 20132013 IEEE International Conference on Systems, Man and Cybernetics (SMC 2013). Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics Piscataway, NJ: IEEE pp. 3014-3019., (10.1109/SMC.2013.514)
Liu, J., Harkin, J., Li, Y. and Maguire, L. 2014. Online traffic-aware fault detection for networks-on-chip. Journal of Parallel and Distributed Computing 74(1), pp. 1984-1993. (10.1016/j.jpdc.2013.09.001)
McDonald, S., Coleman, S., McGinnity, T. M. and Li, Y. 2014. A hybrid forecasting approach using ARIMA models and self-organising fuzzy neural networks for capital markets. Presented at: 2013 International Joint Conference on Neural Networks (IJCNN), Dallas, TX. USA, 4-9 August 2013The 2013 International Joint Conference on Neural Networks (IJCNN). IEEE pp. 1-7., (10.1109/IJCNN.2013.6706965)

2013

Raza, H., Prasad, G. and Li, Y. 2013. EWMA based two-stage dataset shift-detection in non-stationary environments. Presented at: AIAI 2013, Paphos, Cyprus, 30 Sep - 2 Oct 2013 Presented at Papadopoulos, H. et al. eds.Artificial Intelligence Applications and Innovations, Vol. 412. IFIP Advances in Information and Communication Technology Berlin, Heidelberg: Springer pp. 625-635., (10.1007/978-3-642-41142-7_63)
Goel, G., Maguire, L., Li, Y. and McLoone, S. 2013. Evaluation of sampling methods for learning from imbalanced data. Presented at: ICIC 2013, Nanning, China, 28-31 Jul 2013 Presented at Huang, D. et al. eds.Intelligent Computing Theories, Vol. 7995. Information Systems and Applications, incl. Internet/Web, and HCI Berlin, Heidelberg: Springer pp. 392-401., (10.1007/978-3-642-39479-9_47)

2012

Ding, X., Li, Y., Belatreche, A. and Maguire, L. 2012. Constructing minimum volume surfaces using level set methods for novelty detection. Presented at: IJCNN 2012 International Join Conference on Neural Networks, Brisbane, QLD, Australia, 10-15 June 2012The 2012 International Joint Conference on Neural Networks (IJCNN). IEEE International Joint Conference on Neural Networks (IJCNN) IEEE pp. 1-6., (10.1109/IJCNN.2012.6252804)

2011

Zhang, K., Li, Y., Scarf, P. and Ball, A. 2011. Feature selection for high-dimensional machinery fault diagnosis data using multiple models and Radial Basis Function networks. Neurocomputing 74(17), pp. 2941-2952. (10.1016/j.neucom.2011.03.043)
Zhang, K., Ball, A. D., Li, Y. and Gu, F. 2011. A novel feature selection algorithm for high-dimensional condition monitoring data. International Journal of Condition Monitoring 1(1), pp. 33-43. (10.1784/204764211798089075)
Li, Y. and Maguire, L. 2011. Selecting critical patterns based on local geometrical and statistical information. IEEE Transactions on Pattern Analysis and Machine Intelligence 33(6), pp. 1189-1201. (10.1109/TPAMI.2010.188)
Li, Y. 2011. Selecting training points for one-class support vector machines. Pattern recognition letters 32(11), pp. 1517-1522. (10.1016/j.patrec.2011.04.013)

2009

Li, Y. 2009. A surface representation approach for novelty detection. Presented at: International Conference on Information and Automation 2008, Changsha, China, 20-23 June 20082008 International Conference on Information and Automation. IEEE pp. 1464-1468., (10.1109/ICINFA.2008.4608233)

2007

Zhang, K., Ball, A., Gu, F. and Li, Y. 2007. A hybrid model with a weighted voting scheme for feature selection in machinery condition monitoring. Presented at: IEEE International Conference on Automation Science and Engineering 2007, Scottsdale, AZ, United States, 22-25 September 20072007 IEEE International Conference on Automation Science and Engineering. IEEE pp. 174-179., (10.1109/COASE.2007.4341697)

2006

Ben Sasi, A. Y., Gu, F., Li, Y. and Ball, A. D. 2006. A validated model for the prediction of rotor bar failure in squirrel-cage motors using instantaneous angular speed. Mechanical Systems and Signal Processing 20(7), pp. 1572-1589. (10.1016/j.ymssp.2005.09.010)
Li, Y., McLean, D., Bandar, Z., O'Shea, J. and Crockett, K. 2006. Sentence similarity based on semantic nets and corpus statistics. IEEE Transactions on Knowledge and Data Engineering 18(8), pp. 1138-1150. (10.1109/TKDE.2006.130)

2005

Li, Y., Gu, F., Harris, G., Ball, A., Bennett, N. and Travis, K. 2005. The measurement of instantaneous angular speed. Mechanical Systems and Signal Processing 19(4), pp. 786-805. (10.1016/j.ymssp.2004.04.003)
Gu, F., Yesilyurt, I., Li, Y., Harris, G. and Ball, A. 2005. An investigation of the effects of measurement noise in the use of instantaneous angular speed for machine diagnosis. Mechanical Systems and Signal Processing 20(6), pp. 1444-1460. (10.1016/j.ymssp.2005.02.001)

2003

Li, Y., Bandar, Z. and McLean, D. 2003. An approach for measuring semantic similarity between words using multiple information sources. IEEE Transactions on Knowledge and Data Engineering 15(4), pp. 871-882. (10.1109/TKDE.2003.1209005)

2002

Li, Y., Bandar, Z. and Mclean, D. 2002. Measuring semantic similarity between words using lexical knowledge and neural networks. Presented at: 3rd International Conference on Intelligent Data Engineering and Automated Learning — IDEAL 2002, Manchester, England, UK, 12-14 August 2002Intelligent Data Engineering and Automated Learning — IDEAL 2002, Vol. 2412. Lecture Notes in Computer Science Berlin and Heidelberg: Springer pp. 111-116., (10.1007/3-540-45675-9_19)
Li, Y., Pont, M. J. and Jones, N. B. 2002. Improving the performance of radial basis function classifiers in condition monitoring and fault diagnosis applications where 'unknown' faults may occur. Pattern Recognition Letters 23(5), pp. 569-577. (10.1016/S0167-8655(01)00133-7)
Li, Y., Pont, M. J. and Barrie Jones, N. 2002. Improving the performance of radial basis function classifiers in condition monitoring and fault diagnosis applications where `unknown' faults may occur. Pattern recognition letters 23(5), pp. 569-577. (10.1016/S0167-8655(01)00133-7)

2001

Li, Y., Pont, M. J., Jones, N. B. and Twiddle, J. 2001. Applying MLP and RBF classifiers in embedded condition monitoring and fault diagnosis systems. Transactions of the Institute of Measurement and Control 23(5), pp. 315-343. (10.1177/014233120102300504)

2000

Li, Y., Pont, M. J., Parikh, C. R. and Jones, N. B. 2000. Comparing the performance of three neural classifiers for use in embedded applications. Presented at: Workshop 99 on Recent Advances in Soft Computing, Leicester, England, 01-02 July 1999 Presented at John, R. and Birkenhead, R. eds.Soft Computing Techniques and Applications. Advances in Soft Computing Physica pp. 34-29.

Dadansoddeg Data a Grŵp Ymchwil Dysgu Peiriant

Mae fy niddordebau ymchwil yn cynnwys:

Dysgu peirianyddol, adnabod patrymau
Canfod newydd-deb, canfod anomali
Gwyddoniaeth data, Data Mawr, cloddio testun
Rhwydweithiau niwral, dysgu dwfn
Cyfrifiadura hyperddimensiwn, pensaernïaeth symbolaidd fector
Monitro cyflwr a phrosesu signal
Dysgu peirianyddol a chymwysiadau AI, e.e., cyllid, technolegau gofal iechyd, gweithgynhyrchu

Cyhoeddiadau a ddewiswyd (mwy o gyhoeddiadau ar Google Scholar).

Hanzhi Wang, Matthias Treder, Derek Jones, David Marshall, Yuhua Li (2023)
"Swyddogaeth colli sgiw ar gyfer cywiro rhagfarn ragfynegol mewn rhagfynegiad oedran yr ymennydd,"
Trafodion IEEE ar Delweddu Meddygol, vol.42, rhif 6, tt. 1577-1589.
Aboozar Taherkhani, Ammar Belatreche, Yuhua Li, Liam Maguire (2018)
"Algorithm dysgu dan oruchwyliaeth ar gyfer dysgu union amseriad pigau lluosog mewn rhwydweithiau niwral spiking multilayer,"
Trafodion IEEE ar Rwydweithiau Nerfol a Systemau Dysgu, cyf. 29, rhif 11, tt. 5394 - 5407.
Yi Cao, Yuhua Li, Sonya Coleman, Ammar Belatreche, Martin McGinnity (2016)
"Canfod masnach ymolchi yn y farchnad ariannol gan ddefnyddio digraffau a rhaglennu deinamig,"
Trafodion IEEE ar Rwydweithiau Nerfol a Systemau Dysgu, cyf. 27, rhif 11, tt. 2351-2363.
Junxiu Liu, Jim Harkin, Yuhua Li, Liam Maguire (2016)
"Rhwydweithiau goddefgar fai-ar-sglodion llwybro gydag edrych bras a graen mân ymlaen"
Trafodion IEEE ar Dylunio â Chymorth Cyfrifiadur o Gylchedau a Systemau Integredig, cyf. 35, rhif 2, tt. 260-273.
Aboozar Taherkhani, Ammar Belatreche, Yuhua Li, Liam Maguire (2015)
"DL-ReSuMe: Dull goruchwylio dysgu oedi sy'n seiliedig ar ddysgu ar gyfer niwronau nyddu,"
Trafodion IEEE ar Rwydweithiau Nerfol a Systemau Dysgu , vol.26, rhif 12, tt. 3137- 3149.
Xuemei Ding, Yuhua Li, Ammar Belatreche, Liam Maguire (2015)
"Canfod newydd-deb gan ddefnyddio dulliau gosod gwastad,"
Trafodion IEEE ar rwydweithiau niwral a systemau dysgu. Cyf. 26, Rhif 3, tt. 576-588.
Yi Cao, Yuhua Li, Sonya Coleman, Ammar Belatreche, Martin McGinnity (2015)
"Model Markov cudd addasol gyda gwladwriaethau annormal ar gyfer canfod trin prisiau,"
Trafodion IEEE ar Rwydweithiau Nerfol a Systemau Dysgu, vol.26, rhif 2, tt. 318-330.
Haider Raza, Girijesh Prasad, Yuhua Li (2015)
"Dulliau canfod sifft sy'n seiliedig ar fodel EWMA ar gyfer canfod newidiadau cyd-gyfeirio mewn amgylcheddau nad ydynt yn sefydlog,"
Cydnabyddiaeth Patrwm, cyf. 48, rhif 3, tt. 659-669.
Yuhua Li, Liam Maguire (2011)
"Dewis patrymau beirniadol yn seiliedig ar wybodaeth geometrig ac ystadegol leol,"
Trafodion IEEE ar Ddadansoddi Patrymau a Chudd-wybodaeth Peiriant, cyf. 33, rhif 6, tt. 1189-1201.
Yuhua Li (2011)
"Dewis pwyntiau hyfforddi ar gyfer peiriannau fector cymorth un dosbarth,"
Llythyrau Adnabod Patrwm, cyf. 32, rhif 11, tt. 1517-1522.
Yuhua Li, David McLean, Zuhair Bandar, James O'Shea, Keeley Crockett. (2006)
"Tebygrwydd brawddegau gan ddefnyddio rhwydi semantig ac ystadegau corpws,"
Trafodion IEEE ar Beirianneg Gwybodaeth a Data, cyf. 18, rhif 8, tt. 1138-1150.
Yuhua Li, Zuhair Bandar, David McLean (2003)
"Dull o fesur tebygrwydd semantig gan ddefnyddio ffynonellau gwybodaeth lluosog,"
Trafodion IEEE ar Beirianneg Gwybodaeth a Data, cyf. 15, rhif 4, tt. 871-882.
Yuhua Li, Michael Pont, Barrie Jones (2002)
"Gwella perfformiad dosbarthwyr swyddogaeth sail rheiddiol mewn cymwysiadau monitro cyflyrau a diagnosis nam lle gall diffygion "anhysbys" ddigwydd.
Llythyrau Adnabod Patrwm, cyf.23, rhif 5, tt. 569-577.

I received a postgraduate certificate in higher education, I am a Fellow of the HEA. I teach:

CMT307 Applied Machine Learning
CMT316 Applications of Machine Learning: Natural Language Processing/Computer Vision
CMT219 Algorithms, Data Structures and Programming
CM1210 Object Oriented Java Programming

Myfyrwyr PhD cyfredol

Fi yw'r goruchwyliwr arweiniol ar gyfer:

Angga Anggoro (gyda Dr Padraig Corcoran, Dr Dennis De Widt)
James Lewis-Cheetham (gyda Dr Federico Liberatore, Yr Athro Qingwei Wang)
Jianqiao Weng (gyda Dr Oktay Karakus)
Mengqi Zhang (gyda'r Athro Krishna Singh, Yr Athro David Marshall, Dr Matthias Treder)
Nawal Alqurashi (gyda Dr Kirill Sidorov)
Ruizhi Zhang (Prifysgol Beihang)

Mae gen i ddiddordeb mewn goruchwylio myfyrwyr PhD ym meysydd:

Dysgu peirianyddol, adnabod patrymau
Gwyddoniaeth data, Data Mawr, cloddio testun
Rhwydweithiau niwral, dysgu dwfn
Cyfrifiadura hyperddimensiwn, pensaernïaeth symbolaidd fector
Dysgu peirianyddol a chymwysiadau AI, e.e. gofal iechyd, seiberddiogelwch, cyllid a pheirianneg

Mae croeso i chi gysylltu â mi (LiY180@cardiff.ac.uk) os oes gennych gefndir academaidd rhagorol ac uchelgais uchel ar gyfer rhagoriaeth ymchwil. Isod ceir enghreifftiau o gynigion prosiect PhD.

Prosiect 1: Dysgu Peiriant ar gyfer Monitro Iechyd a Chanfod Clefydau Cynnar

Mae Unedau Gofal Dwys Newyddenedigol (NICU) yn defnyddio dyfeisiau meddygol a synwyryddion i gasglu data ffisiolegol yn barhaus fel cyfradd curiad y galon, tymheredd ac ECG. Nod y prosiect hwn yw harneisio'r cyfoeth o ddata sydd ar gael yn NICU i ddatblygu atebion effeithiol ar gyfer monitro newydd-anedig a chanfod clefydau yn gynnar, gyda ffocws penodol ar sepsis, sy'n gyfrifol am 26% o farwolaethau newydd-anedig. Gan ddefnyddio technegau dysgu peiriannau, bydd y prosiect yn tynnu nodweddion o ffrydiau data, yn nodi ffactorau risg, ac yn canfod clefyd yn gynnar i gynorthwyo clinigwyr i ddarparu gwell gofal i gleifion NICU. Bydd y prosiect yn cael ei gynnal mewn partneriaeth â'r Ysgol Meddygaeth, gan ddefnyddio data NICU a gesglir o wardiau ysbytai.

Prosiect 2: Cyfrifiadura hyperddimensiwn ar gyfer Cudd-wybodaeth Peiriant mewn Cymwysiadau Cyfyngedig o Adnoddau

Yn ystod y blynyddoedd diwethaf, mae systemau dysgu peiriannau, gan gynnwys dysgu dwfn, wedi cyflawni llwyddiant rhyfeddol mewn amrywiol gymwysiadau. Fodd bynnag, maent yn wynebu heriau sylweddol, fel defnydd o ynni uchel, addasu i amodau newydd, a diffyg data, megis gwerthoedd coll, anghydbwysedd data, a phreifatrwydd data. Er mwyn mynd i'r afael â'r heriau hyn, mae'r ymchwil hon yn harneisio'r datblygiadau diweddaraf mewn Pensaernïaeth Symbolaidd Fector (VSA) neu Gyfrifiadura Hyperddimensiwn (HDC), fframwaith cyfrifiadurol sy'n dod i'r amlwg, i ddatblygu algorithmau ysgafn a all berfformio prosesu gwybyddol yn effeithlon ar gyfer cymwysiadau â chyfyngiad ar adnoddau ar ymyl y rhwydwaith. Mae VSA wedi'i ysbrydoli gan yr ymennydd ac mae'n cynrychioli ac yn trin data mewn gofod fector dimensiwn uchel. Mae ei gynrychiolaeth ddosbarthedig holograffig a thrin gwybodaeth yn gwneud cyfrifiadura'n fwy cadarn i sŵn, graddadwy, ynni-effeithlon, ac mae angen llai o amser a data ar gyfer hyfforddiant a chasgliad. Nod y prosiect hwn yw datblygu algorithmau sy'n effeithlon o ran adnoddau a data i wella ac ategu dysgu peirianyddol presennol a dulliau dysgu dwfn.

Prosiect 3: Dosbarthiad ar-lein gyda dosbarthiadau newydd sy'n dod i'r amlwg

Dim ond dosbarthiadau a ddiffiniwyd ymlaen llaw y gall dulliau dosbarthu safonol ddosbarthu, h.y., maent yn dosbarthu enghraifft newydd yn un (neu luos) o'r dosbarthiadau hysbys. Er enghraifft, er mwyn adeiladu dosbarthydd ar gyfer clefydau anadlol firaol, mae angen i ni hyfforddi'r model dosbarthu ar set ddata gyda dosbarthiadau wedi'u diffinio ymlaen llaw fel MERS a SARS. Ar adeg datblygu model ar gyfer dosbarthu clefydau, mae'r classifier wedi'i hyfforddi ar y data sydd ar gael sy'n cynnwys yn unig, e.e. MERS a SARS. Dim ond afiechydon MERS a SARS y gall classifier o'r fath ddosbarthu, ni fydd yn gallu delio ag ymddangosiad clefydau newydd fel COVID-19 yn y dyfodol. Er mwyn delio ag ymddangosiad dosbarthiadau newydd, mae angen dull newydd i ddysgu dosbarthwr sy'n gallu canfod dosbarthiadau newydd ac addasu'r classifier yn unol â hynny. Mae gan batrwm dysgu classifier o'r fath gyda dosbarthiadau newydd nifer o gymwysiadau, ee, ceir hunan-yrru yn symud mewn sefyllfaoedd traffig newydd, synwyryddion maleisus sy'n delio â mathau newydd o ymosodiadau rhwydwaith, milwyr robotig sy'n llywio mewn tiroedd newydd, ac ati.

Nod y prosiect hwn yw datblygu dull newydd o ddysgu classifier sy'n gallu dosbarthu dosbarthiadau newydd a newydd. Bydd y dull arfaethedig yn mynd i'r afael â dwy brif her: canfod dosbarthiadau sy'n dod i'r amlwg yn effeithiol ac addasu dosbarthiadau mewn pryd yn unig ar gyfer dosbarthiadau newydd. Bydd canfod dosbarth sy'n dod i'r amlwg yn cael ei adeiladu ar y datblygiadau diweddaraf mewn canfod newydd-deb (techneg dysgu peiriant yw canfod newydd-deb sy'n dysgu model sy'n seiliedig ar ddosbarthiadau hysbys yn unig i ganfod achosion sy'n dod o ddosbarth newydd), a bydd addasu mewn pryd yn datblygu strategaeth ddysgu gynyddrannol newydd i integreiddio dosbarthiadau newydd i'r dosbarth presennol. Bydd yr algorithmau datblygedig yn cael eu gwerthuso ar achos defnydd mewn seiberddiogelwch neu'r Rhyngrwyd Pethau (ee, mathau newydd o ymosodiadau rhwydwaith).

Prosiect 4: Esblygiad cysyniad dysgu mewn ffrydiau data
Mewn cymwysiadau sydd ag esblygiad cysyniad, mae cysyniadau newydd yn dod i'r amlwg yn y llif data, ac mae cysyniadau presennol / hysbys yn diflannu dros amser, ee mathau newydd o ymosodiadau mewn rhwydwaith cyfrifiadurol a phynciau newydd o ddiddordeb mewn ffrwd ddata cyfryngau cymdeithasol. Nod y prosiect hwn yw datblygu dulliau newydd ar gyfer mynd i'r afael â mater heriol esblygiad cysyniad er mwyn galluogi'r modelau dysgedig i ddarparu ar gyfer cysyniadau newydd. Bydd yn cyflawni'r amcanion canlynol: modelu cysyniadau hysbys; achosion newydd o ganfod a chronni; cysyniadau newydd canfod ac integreiddio; cysyniadau hen ffasiwn yn ymddeol.

Prosiect 5: Dysgu peiriant eglurhaol ar gyfer sicrhau Rhyngrwyd Pethau (IoT)
Mae Internet of Things (IoT) yn cynnwys pethau, gwasanaethau a rhwydweithiau, mae'n cysylltu dyfeisiau craff, gwrthrychau, anifeiliaid neu bobl cydberthynol i drosglwyddo data dros rwydwaith i wasanaethu pobl yn well. Mae faint o ddata a drosglwyddir gyda systemau IoT yn barhaus, heterogenaidd ac enfawr, sy'n gwneud systemau IoT yn fwy agored i niwed na'r rhwydwaith traddodiadol i weithgareddau maleisus gan ymosodwyr, felly mae diogelwch a phreifatrwydd y rhwydwaith hynod awtomataidd hwn yn her allweddol ar gyfer defnyddio Rhyngrwyd Pethau (IoT). Mae'n gyson yn destun ymosodiadau gwrthwynebus gan gynnwys gwrthod gwasanaeth, jamio, spoofing, clustodi, malware a gollyngiadau preifatrwydd. Mae'r adnoddau cyfyngedig (cyfrifiant, batri a chof) ar ddyfeisiau IoT a faint o ddata a gynhyrchir ac a gyflewyd yn cyfyngu'n ddifrifol ar gymhwysedd mesurau diogelwch presennol i systemau IoT. Hyd yn oed os yw system ddiogelwch yn effeithiol ar adeg ei defnyddio, mae'n dueddol o fethu cyn gynted ag y mae ymosodwyr yn addasu strategaethau craffach i ffoi'r system ac osgoi canfod. Mae dysgu peirianyddol yn offeryn pwysig ar gyfer canfod ymosodiadau gwrthwynebus, ac mae eglurdeb lefel ddynol o ganlyniadau canfod yn parhau i fod yn agored i ymchwil yn diogelwch IoT.

Nod y prosiect hwn yw mynd i'r afael â'r heriau allweddol hyn i sicrhau systemau IoT yn y dyfodol gyda dulliau dysgu peiriannau creadigol trwy Ymchwilio i ddulliau dosbarthu ffrydio data ar gyfer canfod mathau hysbys o ymosodiadau a'u hamrywiolion yn effeithiol yn y dyfodol; Datblygu algorithmau dysgu peiriannau rhatach cyfrifiadurol yn ogystal â chadernid yn erbyn ymosodiadau clustfeinio; Optimeiddio'r polisi dadlwytho mewn amgylcheddau radio deinamig i ddosbarthu'r llwyth cyfrifiadurol gorau dros gwmwl, dyfais ac ymyl; Ymchwilio i dechnegau dysgu peiriant gwrthwynebol i fynd i'r afael â strategaethau newidiol ymosodwyr; Dehongli canlyniadau rhagfynegi i gefnogi dynol i gymryd camau dibynadwy

Goruchwyliaeth gyfredol

Qing Zheng

Contact Details

LiY180@caerdydd.ac.uk
+44 29208 75317
Abacws, Ystafell 4.59, Ffordd Senghennydd, Cathays, Caerdydd, CF24 4AG

ActiveQuote a Chaerdydd yn dod ynghyd

06 January 2021