var presentation = ["Machine Learning<\/word>","Rob<\/word>","Introduction to Machine Learning<\/strong>\nRob<\/strong> Schapire<\/phrase>","What is Machine Learning<\/strong>?<\/phrase>","studies computer<\/word>","stuff<\/word>","complete<\/word>","accurate predictions<\/word>","behave<\/word>","direct experience<\/word>","Machine learning<\/strong> studies computer<\/strong> algorithms for learning\nto do stuff<\/strong>. We might, for instance, be interested in learning to complete<\/strong> a task, or to make accurate predictions<\/strong>, or to behave<\/strong> intelligently. The learning that is being done is always based on some sort\nof observations or data, such as examples (the most common ca\nse in this course), direct experience<\/strong>, or instruction. So in general, machine learning<\/strong> is about learning to do better in the future based on what was experienced in the past.<\/phrase>","automatic methods<\/word>","other words<\/word>","human intervention<\/word>","The emphasis of machine learning<\/strong> is on automatic methods<\/strong>. In other words<\/strong>, the goal is to devise learning algorithms that do the learning automatically without human intervention<\/strong> or assistance. The machine learning<\/strong> paradigm can be viewed as \"programming by example.\"<\/phrase>","Often<\/word>","specific task<\/word>","spam filtering<\/word>","program<\/word>","own program<\/word>","provide<\/word>","core subarea<\/word>","artificial intelligence<\/word>","intelligent system<\/word>","Often<\/strong> we have a specific task<\/strong> in mind, such as spam filtering<\/strong>. But rather than program<\/strong> the computer to solve the task directly, in machine learning<\/strong>, we seek methods by which the computer will come up with its own program<\/strong> based on examples that we provide<\/strong>. Machine learning<\/strong> is a core subarea<\/strong> of artificial intelligence<\/strong>. It is very unlikely that we will be able to build any kind of intelligent system<\/strong> capable of any of the facilities that we\nassociate with intelligence, such as language or vision, without using learning to get there.<\/phrase>","Further<\/word>","consider<\/word>","other fields<\/word>","theoretical computer science<\/word>","These tasks are otherwise simply too difficult to solve. Further<\/strong>, we would not consider<\/strong> a system to be truly intelligent if it were incapable of learning since learning is at the core of\nintelligence. Although a subarea of AI, machine learning<\/strong> also intersects broadly with other fields<\/strong>, especially statistics, but also mathematics, physics, theoretical computer science<\/strong> and more.<\/phrase>","Machine Learning Problems<\/word>","Examples of Machine Learning Problems<\/strong><\/phrase>","many examples<\/word>","classification problems<\/word>","fixed set<\/word>","several examples<\/word>","optical character recognition<\/word>","categorize images<\/word>","handwritten characters<\/word>","face<\/word>","identify<\/word>","email messages<\/word>","news articles<\/word>","spoken language understanding<\/word>","limited domain<\/word>","There are many examples<\/strong> of machine learning problems<\/strong>. Much of this course will focus on classification problems<\/strong> in which the goal is to categorize objects into a fixed set<\/strong> of categories.\nHere are several examples<\/strong>:\n* optical character recognition<\/strong>: categorize images<\/strong> of handwritten characters<\/strong> by the letters represented\n* face<\/strong> detection: find faces in images (or indicate if a face<\/strong> is p\nresent)\n* spam filtering<\/strong>: identify<\/strong> email messages<\/strong> as spam or non-spam\n* topic spotting: categorize news articles<\/strong> (say) as to whether they are about politics, sports, entertainment, etc.\n* spoken language understanding<\/strong>: within the context of a limited domain<\/strong>, determine the meaning of something uttered by a speaker to the extent that it can be classified into one of a fixed set<\/strong> of categories<\/phrase>","medical diagnosis<\/word>","customer segmentation<\/word>","particular promotion<\/word>","fraud detection<\/word>","credit card transactions<\/word>","rain tomorrow<\/word>","last case<\/word>","* medical diagnosis<\/strong>: diagnose a patient as a sufferer or non-sufferer of some disease\n* customer segmentation<\/strong>: predict, for instance, which customers will respond to a particular promotion<\/strong>\n* fraud detection<\/strong>: identify<\/strong> credit card transactions<\/strong> (for instance) which may be fraudulent in nature\n* weather prediction: predict, for instance, whether or not it will rain tomorrow<\/strong> (In this last case<\/strong>, we most likely would actually be more interested in estimating the probability of rain tomorrow<\/strong>.)<\/phrase>","want<\/word>","fixed categories<\/word>","other hand<\/word>","real value<\/word>","wish<\/word>","In classification, we want<\/strong> to categorize objects into fixed categories<\/strong>. In regression, on the other hand<\/strong>, we are trying to predict a real value<\/strong>. For\ninstance, we may wish<\/strong> to predict how much it will rain tomorrow<\/strong>. Or, we might want<\/strong> to predict how much a house will sell for.<\/phrase>","richer learning scenario<\/word>","intelligent decisions<\/word>","learn<\/word>","use<\/word>","stock market<\/word>","treat<\/word>","classification problem<\/word>","regression problem<\/word>","intermediate goals<\/word>","decide<\/word>","final example<\/word>","play<\/word>","A richer learning scenario<\/strong> is one in which the goal is actually to behave<\/strong> intelligently, or to make intelligent decisions<\/strong>. For instance, a robot needs to learn<\/strong> to navigate through its environment without colliding with anything. To use<\/strong> machine learning<\/strong> to make money on\nthe stock market<\/strong>, we might treat<\/strong> investment as a classification problem<\/strong> (will the stock go up or down) or a regression problem<\/strong> (how much will the stock go\nup), or, dispensing with these intermediate goals<\/strong>, we might want<\/strong> the computer to learn<\/strong> directly how to decide<\/strong> to make investments so as to maximize wealth. A final example<\/strong> is game playing where the goal is for the computer to learn<\/strong> to play<\/strong> well through experience.<\/phrase>","Machine Learning Research<\/word>","Goals of Machine Learning Research<\/strong><\/phrase>","primary goal<\/word>","develop<\/word>","general purpose algorithms<\/word>","practical value<\/word>","Such algorithms<\/word>","computer scientists<\/word>","care<\/word>","time<\/word>","space efficiency<\/word>","great deal<\/word>","precious resource<\/word>","learning algorithm<\/word>","The primary goal<\/strong> of machine learning research<\/strong> is to develop<\/strong> general purpose algorithms<\/strong> of practical value<\/strong>. Such algorithms<\/strong> should be efficient. As usual, as computer scientists<\/strong>, we care<\/strong> about time<\/strong> and space efficiency<\/strong>. But in the context of learning, we also care<\/strong> a great deal<\/strong> about another precious resource<\/strong>, namely, the amount of data that is required by the\nlearning algorithm<\/strong>.<\/phrase>","general purpose<\/word>","broad class<\/word>","Learning algorithms should also be as general purpose<\/strong> as possible. We are looking for algorithms that can be easily applied to a broad class<\/strong> of learning problems, such as those\nlisted above.<\/phrase>","primary importance<\/word>","prediction rule<\/word>","prediction rules<\/word>","human experts<\/word>","Of primary importance<\/strong>, we want<\/strong> the result of learning to be a prediction rule<\/strong> that is as accurate as possible in the predictions that it makes.\nOccasionally, we may also be interested in the interpretability of the prediction rules<\/strong> produced by learning. In other words<\/strong>, in some contexts (such\nas medical diagnosis<\/strong>), we want<\/strong> the computer to find prediction rules<\/strong> that are easily understandable by human experts<\/strong>.<\/phrase>","direct programming<\/word>","First<\/word>","machine learning algorithms<\/word>","examine<\/word>","large amounts<\/word>","human expert<\/word>","imprecise impressions<\/word>","small number<\/word>","As mentioned above, machine learning<\/strong> can be thought of as \"programming by example.\"What is the advantage of machine learning<\/strong> over direct programming<\/strong>? First<\/strong>, the results of using machine learning<\/strong> are often<\/strong> more accurate than what can be created through direct programming<\/strong>. The reason is that machine learning algorithms<\/strong> are data driven, and are able\nto examine<\/strong> large amounts<\/strong> of data. On the other hand<\/strong>, a human expert<\/strong> is likely to be guided by imprecise impressions<\/strong> or perhaps an examination of only a\nrelatively small number<\/strong> of\nexamples.<\/phrase>"]; var currentPosition = 0; var totalPositions = 0; var timePerWord = 1500; var timePerPhraseWord = 120 var readAheadPlayerTimer; var autoPlay = true; var setMinimized = false; var wordSpeedOptions = {'Slower':2000,'Moderate':1200,'Fast':800}; var phraseSpeedOptions = {'Slower':280,'Moderate':200,'Fast':80}; var defaultWordSpeedOption = 'Moderate'; var defaultPhraseSpeedOption = 'Moderate'; var logToLTI = false; $(window).load(function(){ totalPositions = presentation.length; if (totalPositions > 0){ $('body').prepend('
'); $('body').prepend('
'); $('#read_ahead_player').prepend('
'); initializeReadAheadControls(); initializeReadAheadSlider(); initializeReadAheadSpeedOptions(); currentReadAheadControls(); resumeReadAheadPlayback(); } }); $(document).keyup(function(e){ switch(e.which) { case 37: pauseReadAheadPresentation(); priorReadAheadPresentationItem(); break; case 39: pauseReadAheadPresentation(); nextReadAheadPresentationItem(); break; default: break; } }); function initializeReadAheadControls(){ var s = ''; s += ''; s += ''; s += ''; s += '
'; s += ''; s += ''; s += ''; s += ''; s += ''; s += ''; s += ''; s += ''; s += ''; s += '
'; $('#read_ahead_player').append(s); } function initializeReadAheadSlider(){ var s = ''; s += '
'; s += '
'; s += '
'; $('#read_ahead_controls').prepend(s); updateReadAheadSliderPosition(); } function initializeReadAheadSpeedOptions(){ var s = '
'; s += '
'; s += ' Word Speed
'; s += '
'; optionCount = 0; for (index in wordSpeedOptions) { optionCount++; } optionWidth = 100 / optionCount; for (index in wordSpeedOptions) { var option = wordSpeedOptions[index]; if (index == defaultWordSpeedOption) { s += ''; updateReadAheadWordSpeed(option); } else { s += ''; } } s += '
'; s += '
'; s += '
'; s += ' Phrase Speed
'; s += '
'; optionCount = 0; for (index in phraseSpeedOptions) { optionCount++; } optionWidth = 100 / optionCount; for (index in phraseSpeedOptions) { var option = phraseSpeedOptions[index]; if (index == defaultWordSpeedOption) { s += ''; updateReadAheadPhraseSpeed(option); } else { s += ''; } } s += '
'; s += '
'; s += '
'; $('#read_ahead_controls').append(s); } function updateReadAheadWordSpeed(microseconds,sender) { microseconds = microseconds != undefined && microseconds > 0 ? microseconds : 0; if (microseconds > 0) { timePerWord = microseconds; } if (sender!=undefined) { $('#slider_control_word_speed').find('a').removeClass('current_speed'); $(sender).addClass('current_speed'); } logReadAheadPlayerAction('updateReadAheadWordSpeed','Presentation',microseconds); } function updateReadAheadPhraseSpeed(microseconds,sender) { microseconds = microseconds != undefined && microseconds > 0 ? microseconds : 0; if (microseconds > 0 ) { timePerPhraseWord = microseconds; } if (sender!=undefined) { $('#slider_control_phrase_speed').find('a').removeClass('current_speed'); $(sender).addClass('current_speed'); } logReadAheadPlayerAction('updateReadAheadPhraseSpeed','Presentation',microseconds); } function updateReadAheadSliderPosition(){ var currentSliderBarWidth = 100 - (100 / totalPositions * (currentPosition + 1)); $('#readAheadSliderBar').css('width',currentSliderBarWidth+'%'); } function currentReadAheadControls(){ $('#read_ahead_player_pause_play').show(); if (currentPosition < totalPositions - 1) { $('#read_ahead_player_next_word').removeClass('disabled'); } else { $('#read_ahead_player_next_word').addClass('disabled'); } if (currentPosition > 0) { $('#read_ahead_player_previous_word').removeClass('disabled'); } else { $('#read_ahead_player_previous_word').addClass('disabled'); } } function restartReadAheadPresentation(){ currentPosition = 0; logReadAheadPlayerAction('restartReadAheadPresentation','Presentation',''); resumeReadAheadPlayback(); } function resumeReadAheadPlayback(){ $('.fa-repeat').removeClass('fa-repeat').addClass('fa-play'); $('#read_ahead_player_pause_play').each(function(){ $(this).find('.fa-play').removeClass('fa-play').addClass('fa-pause'); $(this).off('click'); $(this).attr('onclick','pauseReadAheadPresentation()'); }); logReadAheadPlayerAction('resumeReadAheadPlayback','Presentation',''); playReadAheadPresentation(); } function pauseReadAheadPresentation(){ $('#read_ahead_player_pause_play').each(function(){ $(this).find('.fa-pause').removeClass('fa-pause').addClass('fa-play'); $(this).off('click'); $(this).attr('onclick','resumeReadAheadPlayback()'); }); logReadAheadPlayerAction('pauseReadAheadPresentation','Presentation',''); read_ahead_pause(); } function endReadAheadPresentation(){ $('.fa-play').removeClass('fa-play').addClass('fa-repeat'); $('.fa-pause').removeClass('fa-pause').addClass('fa-repeat'); $('#read_ahead_player_pause_play').each(function(){ $(this).off('click'); $(this).attr('onclick','restartReadAheadPresentation()'); }); logReadAheadPlayerAction('endReadAheadPresentation','Presentation',''); logLTIEndPresentation(); } function playReadAheadPresentation(){ autoPlay=true; currentPosition--; logReadAheadPlayerAction('playReadAheadPresentation','Presentation',''); nextReadAheadPresentationItem(); } function startReadAheadStepper(timeDisplayWord){ clearInterval(readAheadPlayerTimer); readAheadPlayerTimer = setTimeout(function() { read_ahead_step(); },timeDisplayWord); } function stopReadAheadStepper(){ clearInterval(readAheadPlayerTimer); } function read_ahead_step(){ if (autoPlay) { nextReadAheadPresentationItem(); } } function read_ahead_pause(){ autoPlay=false; stopReadAheadStepper(); } function showReadAheadPresentationItem(index){ currentPosition = index; currentPresentationItem = presentation[currentPosition] != undefined ? presentation[currentPosition].trim() : ""; if (currentPresentationItem != "") { currentItemWordCount = presentation[currentPosition].replace("\n","").split(" ").length; if (currentPresentationItem.startsWith("")){ startReadAheadStepper(currentItemWordCount*timePerWord); $('#read_ahead_player_field').html('
'+currentPresentationItem+'
'); logReadAheadPlayerAction('showReadAheadPresentationItem','Word',currentPresentationItem); } else { startReadAheadStepper(currentItemWordCount*timePerPhraseWord); $('#read_ahead_player_field').html('
'+currentPresentationItem.replace("\n","
")+'
'); logReadAheadPlayerAction('showReadAheadPresentationItem','Phrase',''); } } else { if (currentPosition < totalPositions) { nextReadAheadPresentationItem(); } else { endReadAheadPresentation(); read_ahead_minimize(); } } currentReadAheadControls(); updateReadAheadSliderPosition(); saveReadAheadPosition(); } function priorReadAheadPresentationItem(){ stopReadAheadStepper(); logReadAheadPlayerAction('priorReadAheadPresentationItem','Controls',''); showReadAheadPresentationItem(currentPosition-1); } function nextReadAheadPresentationItem(){ stopReadAheadStepper(); logReadAheadPlayerAction('nextReadAheadPresentationItem','Controls',''); showReadAheadPresentationItem(currentPosition+1); } function read_ahead_playAgain(){ currentWordIndex = 0; logReadAheadPlayerAction('read_ahead_playAgain','Controls',''); playReadAheadPresentation(); } function read_ahead_minimize(){ setMinimized = true; logReadAheadPlayerAction('read_ahead_minimize','Controls',''); read_ahead_switch_min_max(); } function read_ahead_maximize(){ setMinimized = false; logReadAheadPlayerAction('read_ahead_maximize','Controls',''); read_ahead_switch_min_max(); } function read_ahead_switch_min_max(forceSwitch){ forceSwitch = forceSwitch != undefined ? forceSwitch : false; if (forceSwitch) { setMinimized = !setMinimized; } if (setMinimized) { $('#read_ahead_player_back').fadeOut(); $('#read_ahead_player').addClass('minimize'); $('.player-icons .fa-stack').removeClass('fa-lg').addClass('fa-sm'); $('.player-icons .fa-minus').removeClass('fa-minus').addClass('fa-expand'); logReadAheadPlayerAction('read_ahead_switch_min_max','Min',''); //$('#read_ahead_player_max_min').each(function(){ // $(this).off('click'); // $(this).attr('onclick','read_ahead_maximize()'); //}); } else { $('#read_ahead_player_back').fadeIn(); $('#read_ahead_player').removeClass('minimize'); $('.player-icons .fa-stack').removeClass('fa-sm').addClass('fa-lg'); $('.player-icons .fa-expand').removeClass('fa-expand').addClass('fa-minus'); logReadAheadPlayerAction('read_ahead_switch_min_max','Max',''); //$('#read_ahead_player_max_min').each(function(){ // $(this).off('click'); // $(this).attr('onclick','read_ahead_minimize()'); //}); } } function saveReadAheadPosition(){ $.ajax({ type:'POST', url:'/ajax/save_user_presentation_position.php', data: { 'id':230, 'position': currentPosition } }) .done(function(results){ }) .fail(function( jqXHR, textStatus ) { alert( "Request failed: " + textStatus ); }); } function logLTIEndPresentation(action) { if (!logToLTI) { return; } $.ajax({ type:'POST', url:'/lti/presentationEnded.php', data: { 'documentId':230, 'userId':0, } }) } function logReadAheadPlayerAction(action,area,details){ var d = new Date(); var t = d.getTime(); action = action != undefined ? action.trim() : 'UNKNOWN'; area = area != undefined ? area.trim() : ''; details = details != undefined ? details : ''; // If we are about to show a new word, increase the number of keywords // reinforced so we can log that for the Dashboard page. if (area == 'Word') { window.keywordCount++; } $.ajax({ type:'POST', url:'/logging/logPlayerAction.php', data: { 'documentId':230, 'userId':0, 'playerAction':action, 'playerArea':area, 'details':details, 'timestamp': t } }) .done(function(results){ }) .fail(function( jqXHR, textStatus ) { alert( "Request failed: " + textStatus ); }); }