Name – Mehul Singhal Registration Number – 16BCE0823 Software lab – 1 Artificialintelligent Home assistance: – Abstract:- A virtual assistant is a software agent thatcan perform tasks or services for an individual. Sometimes the term”chatbot” is used to refer to virtual assistants generally orspecifically those accessed by online chat (or in some cases online chatprograms that are for entertainment and not useful purposes). As of 2017, the capabilities and usage ofvirtual assistants is expanding rapidly, with new products entering the market.An online poll in May 2017 found the most widely used in the US were Apple’sSiri (34%), Google Assistant (19%), Amazon Alexa (6%), and Microsoft Cortana(4%).
Apple and Google have large installed bases of users on smartphones.Microsoft has a large installed base of Windows-based personal computers,smartphones and smart speakers. Alexa has a large install base for smartspeakers.
History:- The first tool enabled to perform digitalspeech recognition was the IBM Shoebox, presented to the general public duringthe 1962 Seattle World’s Fair after its initial market launch in 1961. Thisearly computer, developed almost 20 years before the introduction of the firstIBM Personal Computer in 1981, was able to recognize 16 spoken words and thedigits 0 to 9. The next milestone in the development of voice recognitiontechnology was achieved in the 1970s at the Carnegie Mellon University inPittsburgh, Pennsylvania with substantial support of the United States Departmentof Defense and its DARPA agency. Their tool “Harpy” mastered about1000 words, the vocabulary of a three-year-old. About ten years later the samegroup of scientists developed a system that could analyze not only individualwords but entire word sequences enabled by a Hidden Markov Model. Thus, the earliest virtual assistants, whichapplied speech recognition software were automated attendant and medicaldigital dictation software. In the 1990s digital speech recognition technologybecame a feature of the personal computer with Microsoft, IBM, Philips andLernout & Hauspie fighting for customers.
Much later the market launch ofthe first smartphone IBM Simon in 1994 laid the foundation for smart virtualassistants as we know them today. The first modern digital virtual assistantinstalled on a smartphone was Siri, which was introduced as a feature of theiPhone 4S on October 4, 2011. Apple Inc.developed Siri following the 2010 acquisition of Siri Inc., a spin-off of SRIInternational, which is a research institute financed by DARPA and the UnitedStates Department of Defense. AboutVirtual assistants: – A virtual assistant (VA) is a person whoprovides support services to other businesses from a remote location. The termoriginated in the 1990s as the ability to work virtually due to technologyimprovements, such as high speed Internet, document sharing, and and otheradvancements, made working remotely a reality. Virtual assistants are especially in demand bysolo-preneurs and online businesses that need help, but don’t want to bring onstaff in their location.
However, many small and mid-size businesses usevirtual support, especially in specific tasks such as social media management. Theoretically, a VA can do anything any othersupport staff does, except bring the coffee. (Although when home-deliverycoffee is created, the VA will be able to do that too!). However, virtualsupport duties are not limited to clerical work.
Many VAs provide marketing,web design and other services. A basic list of services include: Calendar managementEmail managementSocial media managementAppointment settingMarketing and PRResearchWritingGraphic creationWebsite managementBookkeepingCustomer supportProject managementTravel bookingCustomer service Some virtual assistants specialize in aspecific skill set. For example, a marketing or PR virtual assistant only doesmarketing or PR work. Other virtual assistants do a variety of duties, butwithin a specific industry. Artificial intelligence (AI, also machineintelligence, MI) is intelligence displayed by machines, in contrast with thenatural intelligence (NI) displayed by humans and other animals. In computerscience AI research is defined as the study of “intelligent agents”:any device that perceives its environment and takes actions that maximize itschance of success at some goal. Colloquially, the term “artificialintelligence” is applied when a machine mimics “cognitive”functions that humans associate with other human minds, such as “learning”and “problem solving”. See glossary of artificial intelligence.
The scope of AI is disputed: as machines becomeincreasingly capable, tasks considered as requiring “intelligence”are often removed from the definition, a phenomenon known as the AI effect,leading to the quip “AI is whatever hasn’t been done yet.” Forinstance, optical character recognition is frequently excluded from”artificial intelligence”, having become a routine technology.Capabilities generally classified as AI as of 2017 include successfullyunderstanding human speech, competing at a high level in strategic game systems(such as chess and Go), autonomous cars, intelligent routing in contentdelivery networks, military simulations, and interpreting complex data,including images and videos. Artificial intelligence was founded as anacademic discipline in 1956, and in the years since has experienced severalwaves of optimism, followed by disappointment and the loss of funding (known asan “AI winter”), followed by new approaches, success and renewedfunding.
For most of its history, AI research has been divided into subfieldsthat often fail to communicate with each other. These sub-fields are based ontechnical considerations, such as particular goals (e.g. “robotics”or “machine learning”), the use of particular tools(“logic” or “neural networks”), or deep philosophicaldifferences. Subfields have also been based on social factors (particularinstitutions or the work of particular researchers). The traditional problems (or goals) of AIresearch include reasoning, knowledge, planning, learning, natural languageprocessing, perception and the ability to move and manipulate objects. Generalintelligence is among the field’s long-term goals. Approaches includestatistical methods, computational intelligence, and traditional symbolic AI.
Many tools are used in AI, including versions of search and mathematicaloptimization, neural networks and methods based on statistics, probability andeconomics. The AI field draws upon computer science, mathematics, psychology,linguistics, philosophy, neuroscience, artificial psychology and many others. The field was founded on the claim that humanintelligence “can be so precisely described that a machine can be made tosimulate it”. This raises philosophical arguments about the nature of themind and the ethics of creating artificial beings endowed with human-likeintelligence, issues which have been explored by myth, fiction and philosophysince antiquity. Some people also consider AI a danger to humanity if itprogresses unabatedly.
Others believe that AI, unlike previous technologicalrevolutions, will create a risk of mass unemployment. In the twenty-first century, AI techniques haveexperienced a resurgence following concurrent advances in computer power, largeamounts of data, and theoretical understanding; and AI techniques have becomean essential part of the technology industry, helping to solve many challengingproblems in computer science. ExistingWork: – APPLESIRIBio: A voice-drivenassistant that talks back to you–invoked by long-pressing the iPhone or iPadhome button–and proactively recommends actions to take. Recently took upresidence on Apple TV and Apple Watch. Notable skills: Easyto access on Apple devices.
Understands natural human language. Knowledgeableabout news, weather, sports, movies, directions, and local businesses.Well-versed in what to watch on TV. Knows how to control some smart homeappliances.Character flaws: Doesn’tknow how to communicate with most other apps and services. Not always as fastas some assistants.Level of humanity: Can’thold an extended conversation, but cracks wise when given the chance. Femalevoice doesn’t sound overly robotic.
Outlook: Siri pavedthe way for modern speech-based assistants, but hasn’t gotten significantlysmarter over the past few years. The lack of an open API means you can’t open asong in Spotify, add a task to Wunderlist, or post a message in Slack, even astie-ins with other apps become table stakes among other virtual assistants.Apple must figure out these types of integrations for Siri to stay relevant;maybe we’ll hear news about them at next month’s WWDC keynote. GOOGLE VOICE SEARCH/GOOGLE NOWBio: Voiceassistant powered by the world’s largest search engine. Also digs through youremail and search history to help you out. Lives on Android devices, iOS, andChrome. Notable skills: Fast.Uncommonly accurate with directions.
Eerily adept at mining your personal datafor flights, packages, reservations, and other useful info. Has some capacity to speak withthird-party apps forcertain tasks, including notes, messages, and music playback.Character flaws: Attemptsat proactivity can sometimes be a nuisance (e.
g., sports scores for teams youdon’t care about, directions home from familiar places). No hands in the smarthome business. Third-party app integrations seem to have stalled.Level of humanity: None.Averse to conversation and doesn’t even have a name, aside from “Google.”Outlook: Google’svast troves of personal data and search engine knowledge should in theory allowit to dominate the AI business, yet Google hasn’t quite figured out how to turnthose advantages into an assistant that truly gets you.
For now, Google Now andvoice search are capable rivals to Siri, but haven’t reached the next level.AMAZON ALEXABio: Voice-activatedassistant that lives on Amazon audio gear (Echo, Echo Dot, Tap) and Fire TVboxes and is making its way to other connected devices such as alarm clocks andpet feeders. Notable skills: Streamsmusic and reads news from multiple sources. Provides weather, traffic, andother info, and controls a growing number of smart home devices.
Allows voicepurchases for Amazon Prime items and even lets you order a pizza. Open API letsany app or service tie into it.Character flaws: Houseboundwith no smartphone integration. May make you wonder if you’re nothing but areceptacle for Amazon goods and services.Level of humanity: Employsa touch of banter with tricky questions, but is quick to guide you back tobusiness.
(“Alexa, what should I do with my life?” “You should write thatnovel. Amazon Kindle Self-Publishing will help you when you’re done.”)Outlook: Apple andGoogle should be terrified of Alexa, which is quickly gaining developermomentum and is now leaping onto new, non-Amazon devices.
Still, Amazon doesn’thave its own smartphone platform–anymore–which means Siri and Google’s assistant have anadvantage on the one device that matters most to people. MICROSOFT CORTANABio: Voice- andtext-based virtual assistant that’s available on Windows, iOS, and Android.Combines proactive knowledge with answers to queries. Might someday help stop aliens from extinguishingall intergalactic life. Notable skills: Handlesreminders and calendar appointments, tracks packages, sets alarms, and tapsinto Bing for sports, weather, and other information. Hooks into some Windowsapps, and has recently started talking to other bots in Skype.Character flaws: Feelsmost at home on Windows, the platform that app developers–and, arguably,users–care the least about. Has fewer capabilities and is harder to access oniOS and Android.
Level of humanity: Lovesjokes, especially corny ones, and has a long list of wisecracks athand forgeneric questions. Will also quote Shakespeare.Outlook: Afteryears as an also-ran behind Siri and Google, Cortana has become a much moreambitious chatbot.
Microsoft wants its virtual assistant to serve as a master intelligence for allkinds of other bots,guiding you through travel plans, meetings, to-do lists, and more, and to bedeeply integrated with other Microsoft products such as Office. The goal is toredefine computing in the post-PCera, but it’s tooearly to tell whether the company will succeed. FACEBOOKMBio: Partartificial intelligence, part human-powered service, still in development. M isto be a text-based assistant within Facebook Messenger that helps get thingsdone.
Notable skills: Attemptsto do anything you might ask it to do.Character flaws: Doesn’tactually exist as a consumer product, and is a long way from getting there. Only a small number of people in SanFrancisco have access.
Level of humanity: Extremelyhigh, as M relies in large part on real humans to answer queries. Thehope, according to Wired, is that thousands of these helpers willtrain M to work on its own over time.Outlook: At themoment, M is little more than vaporware. But given Facebook’s interest in chatbots as a whole, don’t count out M’s eventual arrival as asuperintelligence.SOUNDHOUND HOUNDBio: Voiceassistant app for iOS and Android. A related service called Houndify will letthird-party developers add voice to their own devices and services. Notable skills: Impressiveunderstanding of complex requests such as “Show me coffee shops within fivemiles that aren’t Starbucks.
” Ties into some third-party services such as Yelp,Uber, and Expedia.Character flaws: Connectionsto third-party apps are limited, and no shortcut exists to open the app on iOSand Android.Level of humanity: Isn’tmuch for idle chitchat, but knows how to respond to follow-up questions afteran initial query.
Outlook: One getsthe feeling that Hound’s mobile apps are just a showcase for the Houndifyservice that SoundHound is hoping to sell to other companies. If it succeeds,you probably won’t even recognize it’s there. VIVBio: Virtualassistant from the inventors of Siri. Not available yet, but intended to run onall kinds of computing devices. Notable skills: Viv’sclaim to fame is that it can interpret complex questions such as “Will it bewarmer than 70 degrees near the Golden Gate Bridge after 5 p.
m. the day aftertomorrow?” Tie-ins with third-party apps like Venmo are in the works.Character flaws: Littleproof that it actually functions as advertised outside of prepared demos.
Level of humanity: Appearsto favor visuals and actual information over descriptive feedback. Capacity forbanter is unclear.Outlook: Viv hasreceived plenty of hype from the tech press, as its natural language skillsmake for an impressive demo. But until the startup announces when it’ll launchand on what devices, it’s best to remain a teeny bit skeptical of itsworld-changing claims.OZLOBio: An AI whosesole purpose, at least for now, is to help you find things to eat and drink.Available to a limited number of early sign-ups. Notable skills: Findsand internalizes data from multiple sources such as Yelp and Foursquare,pulling it all into slick informational cards. Tries to be conversational byoffering and understanding follow-up questions, such as “which ones are opennow?” and “what’s on the menu?”Character flaws: Limitedutility, at least until Ozlo’s makers start adding more capabilities.
Heavyreliance on users to train the AI.Level of humanity: Appearsto avoid human pleasantries beyond a brief greeting by name.Outlook: Ozlowouldn’t be much different from the long list of other single-purpose chatbotsif it wasn’t promising to be something bigger. Its ability to string togetherdifferent data sources in a single query is unique, but it’s unclear if the appcan fulfill the potential that its creators are promising. And unless Ozlo hasa business plan that involves more than just a downloadable app, it may havetrouble getting the training data it apparently needs to succeed. X.AIBio: One of severalsingle-purpose virtual assistants. Exists solely via email to schedule meetingson your behalf.
Notable skills: Knowsyour schedule and preferences, handles the legwork of corresponding with otherparties.Character flaws: Reliesheavily on humans to verify the vast majority of calendar data from emails thatthe virtual assistant, “Amy,” generates, according to Bloomberg.Level of humanity: Unsurprisingly,has been praised for its humanlike capabilities and tone.Outlook: Highlyfocused intelligent agents like X.ai will be great if they ever become smartenough to operate autonomously.
Then again, people who don’t mind theappearance of having an assistant for meetings might be able to afford realassistants.SPEAKTOIT ASSISTANT.AIBio: One of manySiri knockoffs, for lack of a more charitable term.
A search for “Siri” in theapp store brings up plenty of others, such as Voice Commands, Voice Secretary,and Assistant. Notable skills: Littleto speak of beyond Siri, but it can learn custom voice commands to activate itsexisting list of skills.Character flaws: Notas useful as the virtual assistant your phone comes with, and not as easilyaccessible.Level of humanity: Soundspretty robotic, but presents itself as a drawing of a human secretary, whosegender and appearance are customizable.Outlook: Some ofthese Siri-alikes seem like holdovers from when not all iPhone models supportedApple’s own virtual assistant and needed a stand-in. In any case, their makersmay have realized the idea isn’t a winning one. SpeakToIt, for instance, haspivoted to a set of tools that help developers make their ownchatbots.
Algorithms:- Decision Trees: A decision tree is adecision support tool that uses a tree-like graph or model of decisions andtheir possible consequences, including chance-event outcomes, resource costs,and utility. Take a look at the image to get a sense of how it looks like.DecisionTreeFrom a business decision pointof view, a decision tree is the minimum number of yes/no questions that one hasto ask, to assess the probability of making a correct decision, most of thetime. As a method, it allows you to approach the problem in a structured andsystematic way to arrive at a logical conclusion.2.
Naive Bayes Classification: Naive Bayes classifiersare a family of simple probabilistic classifiers based on applying Bayes’theorem with strong (naive) independence assumptions between the features. Thefeatured image is the equation?—?with P(A|B) is posterior probability, P(B|A)is likelihood, P(A) is class prior probability, and P(B) is predictor priorprobability. NaiveBayes ClassificationSome of real world examplesare:· To mark an email as spam or notspam· Classify a news article abouttechnology, politics, or sports· Check a piece of textexpressing positive emotions, or negative emotions?· Used for face recognitionsoftware.
3. Ordinary Least Squares Regression: If you know statistics,you probably have heard of linear regression before. Least squares is a methodfor performing linear regression. You can think of linear regression as thetask of fitting a straight line through a set of points. There are multiplepossible strategies to do this, and “ordinary least squares” strategy go likethis?—?You can draw a line, and then for each of the data points, measure thevertical distance between the point and the line, and add these up; the fittedline would be the one where this sum of distances is as small as possible. OrdinaryLeast Squares RegressionLinear refers the kind of modelyou are using to fit the data, while least squares refers to the kind of errormetric you are minimizing over.4. Logistic Regression: Logistic regression is apowerful statistical way of modeling a binomial outcome with one or moreexplanatory variables.
It measures the relationship between the categoricaldependent variable and one or more independent variables by estimatingprobabilities using a logistic function, which is the cumulative logisticdistribution. LogisticRegressionIn general, regressions can beused in real-world applications such as:· Credit Scoring· Measuring the success rates ofmarketing campaigns· Predicting the revenues of acertain product· Is there going to be anearthquake on a particular day?5. Support Vector Machines: SVM is binaryclassification algorithm. Given a set of points of 2 types in N dimensionalplace, SVM generates a (N?—?1) dimensional hyperplane to separate those pointsinto 2 groups. Say you have some points of 2 types in a paper which arelinearly separable.
SVM will find a straight line which separates those pointsinto 2 types and situated as far as possible from all those points. SupportVector MachineIn terms of scale, some of thebiggest problems that have been solved using SVMs (with suitably modifiedimplementations) are display advertising, human splice site recognition,image-based gender detection, large-scale image classificationPseudocode: – For an agent learning from experience to act onsome environment:1. Try something (from a listof available actions for the current situation)2. If the outcome is good, tryit more frequently in the future when in the same situation3. Otherwise, try it lessfrequently in the future when in the same situation4. Go to 1 Futureof AI : – Technology moves at breakneck speed, and we now have more power in ourpockets than we had in our homes in the 1990s. Artificial intelligence (AI) hasbeen a fascinating concept of science fiction for decades, but many researchersthink we’re finally getting close to making AI a reality. NPR notes that in thelast few years, scientists have made breakthroughs in “machine learning,” usingneural networks, which mimic the processes of real neurons.
This is a type of “deep learning” that allows machines to processinformation for themselves on a very sophisticated level, allowing them toperform complex functions like facial recognition. Big data is speeding up theAI development process, and we may be seeing more integration of AI technologyin our everyday lives relatively soon. While much of this technology is stillfairly rudimentary at the moment, we can expect sophisticated AI to one daysignificantly impact our everyday lives. Here are 6 ways AI might affect us inthe future. WhatI conveyed and what you achieved as conclusion: – Since AI is a very big and humongous topic Iwould like to do more research and learn about current trends.