生物信息学-第3章，人工智能导论
生物信息学- 第3章，人工智能导论
人工智能技术人工智能涉及计算机算法的开发，以执行通常与人类智能相关的任务。人工智能包括但不限于机器学习、表征学习、深度学习和自然语言处理。无论具体技术如何，这些技术在医学上的总体目标是使用计算机算法从数据中发现相关信息并协助临床决策AlapproachStudydesignSetupInputDutasets hatAigorithms thatStudies thatResearch that isare simplisticroquire alot offocusonsetupfor Alunimodalanddevelopmentexpertversus-humanConventionaiannotationshomogeneousandrerospectvecomparisonworkflowvalidationStudies thatResearchthatDatasetsthaAigorithmsthatfoous on clinicalwre real-worid,workwitnis.set upforlimited labeledAi-humandeploymentmultimodalandNewheterogeneousdiutausingnovelcollaborationopportunitiessee-supervisecAAandsemi-supervisedapproacheRajpurkar, P., Chen,E., Banerjee, O. & Topol, E. J. (2022). AI in health and medicine.Nature medicine,28(1), 31-38
人工智能技术 人工智能涉及计算机算法的开发，以执行通常与人类智能相关的任务。 人工智能包括但不限于机器学习、表征学习、深度学习和自然语言处 理。无论具体技术如何，这些技术在医学上的总体目标是使用计算机 算法从数据中发现相关信息并协助临床决策。 Rajpurkar, P., Chen, E., Banerjee, O., & Topol, E. J. (2022). AI in health and medicine. Nature medicine, 28(1), 31-38
AI简介
AI 简介
ArtificialIntelligenceARTIFICIALINTELLIGENCEEarlyartificial intelligenceMACHINEstirsexcitement.LEARNINGMachine learning beginsDEEPto flourish.LEARNINGDeep.learningbreakthroughsdriveAlboom.1960's1970's1980's1990's2000s2010's1950'sSince an early flush ofoptimism in the1950s,5maller subsets ofartificial intelligence-firstmachine learning.thendeeplearning,asubsetofmachinelearning-havecreatedeverlargerdisruptions
Artificial Intelligence
Intelligence智能What is intelligence?脑机接口第四关沙康Wikipedia : Intelligence has been defined in many ways to include thecapacityfor logic,understanding,self-awareness,learning,emotionalknowledge,reasoning.planning,creativity,andproblem.solving.Itcanbemoregenerallydescribedastheabilitytoperceiveorinferinformation,andtoretain it asknowledgetobeapplied towardsadaptivebehaviors withinan品还有二主球environmentorcontext（包括：逻辑，理解，自我意识，学习，情感，溯源，计划，创造，问题解决）自动驾驶机器人什么是人工智能ArtificialIntelligence？，对照智能的定义，在某种程度上接近或者超过人AI生成的能力常见情景：预测，判断，语言，下棋，推理，交流，数学推导如何把智能活动数学化语言与智能的关系
Intelligence 智能 What is intelligence? Wikipedia : Intelligence has been defined in many ways to includethe capacity for logic, understanding, self-awareness, learning, emotional knowledge, reasoning, planning, creativity, and problem solving. It can be more generally described as the ability to perceive or infer information, and to retain it as knowledge to be applied towards adaptive behaviors within an environment or context. 什么是人工智能 Artificial Intelligence? ▪ 对照智能的定义，在某种程度上接近或者超过人 的能力 ▪ 常见情景：预测，判断，语言，下棋，推理，交 流，数学推导，. . ▪ 如何把智能活动数学化 语言与智能的关系 脑机接口 机器人 AI生成 （包括：逻辑，理解，自我意识，学习，情感，溯源，计划，创造，问题解决） 自动驾驶
人工智能历史1956DartmouthConference:TheFoundingFathers of Al标志性事件：1956年的Dartmouth会议，十人左右参加，历时两个月。当时创立的计算机系通常分三个领域：理论，系统，AIAI领域显示度较高的项目：下棋。从1957年开始，每十年都预言十年内机器能下赢国际象棋世界冠军。一直等了40年，到1997才真正实现。量变到质变：1995年卡斯帕罗夫批评计算机下棋没有悟性/直觉(Insights）：1997年失败的时候还在猜是不是计算机作弊（有人类象棋大师群在背后集体决策）从这里可以推论悟性/直觉指的是什么机器会思考吗？E乔姆斯基回答“潜艇会游泳吗？
人工智能历史 标志性事件：1956年的 Dartmouth 会议，十人左右参加，历时两个月。 当时创立的计算机系通常分三个领域：理论，系统，AI。 AI领域显示度较高的项目：下棋。从1957年开始，每十年都预言十年内 机器能下赢国际象棋世界冠军。一直等了40年，到1997才真正实现。 • 量变到质变：1995年卡斯帕罗夫批评计算机下棋没有悟性/直觉 (Insights)；1997年失败的时候还在猜是不是计算机作弊（有人类象棋大 师群在背后集体决策） • 从这里可以推论悟性/直觉指的是什么 机器会思考吗？ 乔姆斯基回答 “潜艇会游泳吗？
思考题，在chatGPT可以进行类人创作和问答之后，还有什么是智能体/大模型理论上做不到的？23
▪ 在chatGPT 可以进行类人创作和问答之后，还有什么是智能体/大模型 理论上做不到的？ ▪ 1 ▪ 2 ▪ 3 思考题
Automated Theorem Provingsm- Russell & Whitehead, 1910s, Principia Mathematica, derive allmathematical truth using axioms and inference rules of formal*5443.i.a.Bel.D:anBmA.=.aupe2logic.Dem.F.*5426.ofi.awta.p=ty.D:aupe2.#.ay.[51-231]=.t'anty=A“Logic Theorist"could provea[*13-12](1)=.anp-A.(1).*11:11-35.)subset in Principia Mathematica.(2)Fi.(az,y).a-ta.Bmty.D:aue2.=.anp=AF.(2).*11-54.*52-1.DF.PropFrom this proposition it will follow, when arithmetical addition has been1959, Hao Wang provedalldefined, that 1+1=2.theorems in PrincipiaMathematicaon an IBM 704.1976,fourcolortheorem.- 2016, Boolean Pythagorean Triple Problem, 200TB. There are10^2355 possible solutions, compare with 10^80 atoms in theuniversity and 10^18 seconds since the big bang
▪ Russell & Whitehead, 1910s, Principia Mathematica, derive all mathematical truth using axioms and inference rules of formal logic. ▪ “Logic Theorist” could prove a subset in PrincipiaMathematica. ▪ 1959, Hao Wang provedall theorems in Principia Mathematica on an IBM 704. ▪ 1976, four color theorem. ▪ 2016, Boolean Pythagorean Triple Problem, 200TB. There are 10^2355 possible solutions, compare with 10^80 atoms in the university and 10^18 seconds since the big bang. Automated Theorem Proving
ExpertSystem1960s,thefirstexpert system,DENTRAL1970s,MYCIN,identify bacteriacausing severeinfectionsandrecommendantibiotics1980s, XCON, assist in the ordering ofDEC's VAXcomputersystems by automatically selecting the computer systemcomponentsbasedonthecustomer'srequirementsNow still usedin credit investigation,risk management, businessrules engine,etc
▪ 1960s, the first expert system,DENTRAL. ▪ 1970s, MYCIN, identify bacteria causing severe infectionsand recommend antibiotics. ▪ 1980s, XCON, assist in the ordering of DEC's VAXcomputer systems by automatically selecting the computer system components based on the customer's requirements. ▪ Now still used in credit investigation, risk management, business rules engine, etc. Expert System
Knowledge RepresentationLogics:(Vx)Man(x)Mortal(x) & Man(Socrates) → Mortal(Socrates)WordNetSearch-3.1WordNethome fageClossary-HeipWord to.search for:.HamanSearch.WordietDrsplayOptions(Seiect opban techange)ChangShowWord(leccal)reltotKrSShow:Synsetisepiayopitioms.torsensn.(glossj"anexamgiesertoncoWordNet:NounS(n)momo.man,humanboing.human (any living orextinctmemberot thefamtyaecharactenzedtvsupernorntellonce.arfculalesoeethanderectcamageAdjective(adi)human (charactoristicofhumanity)humannature(adhuman (relatingtoaperson)zheexperimentwasconductedon.6monkeysand2humans(ad)human (navinghumanform orattributes as opposedto thoseofanimais ordivinebengs)humanbeings"hehuwnanbodyhmankndness"hcmantfraityConceptualgraph,SemanticwebKnowledgegraph
▪ Logics: ∀𝑥 𝑀𝑎𝑛 𝑥 ⊃ 𝑀𝑜𝑟𝑡𝑎𝑙 𝑥 & 𝑀𝑎𝑛(𝑆𝑜𝑐𝑟𝑎𝑡𝑒𝑠) → 𝑀𝑜𝑟𝑡𝑎𝑙(𝑆𝑜𝑐𝑟𝑎𝑡𝑒𝑠) ▪ WordNet: ▪ Conceptual graph, Semanticweb ▪ Knowledge graph Knowledge Representation