hunch_net hunch_net-2009 hunch_net-2009-352 knowledge-graph by maker-knowledge-mining
Source: html
Introduction: I recently had fun discussions with both Vikash Mansinghka and Thomas Breuel about approaching AI with machine learning. The general interest in taking a crack at AI with machine learning seems to be rising on many fronts including DARPA . As a matter of history, there was a great deal of interest in AI which died down before I began research. There remain many projects and conferences spawned in this earlier AI wave, as well as a good bit of experience about what did not work, or at least did not work yet. Here are a few examples of failure modes that people seem to run into: Supply/Product confusion . Sometimes we think “Intelligences use X, so I’ll create X and have an Intelligence.” An example of this is the Cyc Project which inspires some people as “intelligences use ontologies, so I’ll create an ontology and a system using it to have an Intelligence.” The flaw here is that Intelligences create ontologies, which they use, and without the ability to create ont
sentIndex sentText sentNum sentScore
1 As a matter of history, there was a great deal of interest in AI which died down before I began research. [sent-3, score-0.128]
2 Sometimes we think “Intelligences use X, so I’ll create X and have an Intelligence. [sent-6, score-0.135]
3 ” An example of this is the Cyc Project which inspires some people as “intelligences use ontologies, so I’ll create an ontology and a system using it to have an Intelligence. [sent-7, score-0.198]
4 ” The flaw here is that Intelligences create ontologies, which they use, and without the ability to create ontologies you don’t have an Intelligence. [sent-8, score-0.605]
5 If we are unlucky, it fails to even be partially useful, because the format is unnatural for the internal representations of an Intelligence. [sent-10, score-0.176]
6 If you asked the people working on them, they might agree that uncertainty was an important but secondary concern to be solved after the main problem. [sent-13, score-0.557]
7 Unfortunately, it seems that uncertainty is a primary concern in practice. [sent-14, score-0.418]
8 One example of this is blocks world where a system for planning how to rearrange blocks on a table might easily fail in practice because the robot fails to grab a block properly. [sent-15, score-0.612]
9 Many people think of uncertainty as a second order concern, because they don’t experience uncertainty in their daily lives. [sent-16, score-0.847]
10 I believe this is incorrect—a mental illusion due to the effect that focusing attention on a specific subject implies reducing uncertainty on that subject. [sent-17, score-0.312]
11 More generally, because any Intelligence is a small part of the world, the ability of any intelligence to perceive, understand, and manipulate the world is inherently limited, requiring the ability to deal with uncertainty. [sent-18, score-0.632]
12 Some people try to create an intelligence without reference to efficient computation. [sent-22, score-0.518]
13 The algorithm is very difficult to deploy in practice because there were no computational constraints other than computability designed into it’s creation. [sent-24, score-0.336]
14 There was a time when some people thought, “If we could just get a program that mastered chess so well it could beat the best humans, we will learn enough about AI to create an AI. [sent-27, score-0.198]
15 Here A might be nearest neighbors, decision trees, two-layer neural networks, support vector machines, nonparametric statistics, nonparametric Bayes, or something else. [sent-34, score-0.25]
16 Solving AI is undeniably hard, as evidenced by the amount of time spent on it, and the set of approaches which haven’t succeeded. [sent-37, score-0.175]
17 The first is that there is, or soon will be sufficient computation available, unlike the last time. [sent-39, score-0.157]
18 The second is that the machine learning approach fails well, because there are industrial uses for machine learning. [sent-40, score-0.201]
19 The machine learning approach to AI has this goodness property, unlike many other approaches, which partially explains why the ML approach is successful despite “failing” so far to achieve AI. [sent-45, score-0.138]
20 Given this, a fair strategy seems to be first mastering one strategy, and then incorporating others, always checking that that incorporation properly addresses real world problems. [sent-47, score-0.244]
wordName wordTfidf (topN-words)
[('ai', 0.348), ('uncertainty', 0.312), ('intelligence', 0.252), ('intelligences', 0.195), ('ontologies', 0.195), ('create', 0.135), ('constraints', 0.131), ('blocks', 0.13), ('cyc', 0.13), ('imitate', 0.13), ('fails', 0.11), ('world', 0.108), ('undeniably', 0.107), ('concern', 0.106), ('humans', 0.104), ('second', 0.091), ('computational', 0.089), ('nonparametric', 0.087), ('computation', 0.085), ('ml', 0.078), ('strategy', 0.078), ('might', 0.076), ('unlike', 0.072), ('ability', 0.072), ('deal', 0.07), ('property', 0.07), ('experience', 0.069), ('approaches', 0.068), ('without', 0.068), ('programs', 0.067), ('partially', 0.066), ('people', 0.063), ('statistics', 0.062), ('everything', 0.062), ('substantial', 0.061), ('chunking', 0.058), ('upside', 0.058), ('unlucky', 0.058), ('mastering', 0.058), ('breuel', 0.058), ('aixi', 0.058), ('asymptopia', 0.058), ('breath', 0.058), ('computability', 0.058), ('decomposition', 0.058), ('deploy', 0.058), ('died', 0.058), ('fool', 0.058), ('grab', 0.058), ('manipulate', 0.058)]
simIndex simValue blogId blogTitle
same-blog 1 1.0000002 352 hunch net-2009-05-06-Machine Learning to AI
Introduction: I recently had fun discussions with both Vikash Mansinghka and Thomas Breuel about approaching AI with machine learning. The general interest in taking a crack at AI with machine learning seems to be rising on many fronts including DARPA . As a matter of history, there was a great deal of interest in AI which died down before I began research. There remain many projects and conferences spawned in this earlier AI wave, as well as a good bit of experience about what did not work, or at least did not work yet. Here are a few examples of failure modes that people seem to run into: Supply/Product confusion . Sometimes we think “Intelligences use X, so I’ll create X and have an Intelligence.” An example of this is the Cyc Project which inspires some people as “intelligences use ontologies, so I’ll create an ontology and a system using it to have an Intelligence.” The flaw here is that Intelligences create ontologies, which they use, and without the ability to create ont
2 0.28121743 380 hunch net-2009-11-29-AI Safety
Introduction: Dan Reeves introduced me to Michael Vassar who ran the Singularity Summit and educated me a bit on the subject of AI safety which the Singularity Institute has small grants for . I still believe that interstellar space travel is necessary for long term civilization survival, and the AI is necessary for interstellar space travel . On these grounds alone, we could judge that developing AI is much more safe than not. Nevertheless, there is a basic reasonable fear, as expressed by some commenters, that AI could go bad. A basic scenario starts with someone inventing an AI and telling it to make as much money as possible. The AI promptly starts trading in various markets to make money. To improve, it crafts a virus that takes over most of the world’s computers using it as a surveillance network so that it can always make the right decision. The AI also branches out into any form of distance work, taking over the entire outsourcing process for all jobs that are entirely di
3 0.27320382 353 hunch net-2009-05-08-Computability in Artificial Intelligence
Introduction: Normally I do not blog, but John kindly invited me to do so. Since computability issues play a major role in Artificial Intelligence and Machine Learning, I would like to take the opportunity to comment on that and raise some questions. The general attitude is that AI is about finding efficient smart algorithms. For large parts of machine learning, the same attitude is not too dangerous. If you want to concentrate on conceptual problems, simply become a statistician. There is no analogous escape for modern research on AI (as opposed to GOFAI rooted in logic). Let me show by analogy why limiting research to computational questions is bad for any field. Except in computer science, computational aspects play little role in the development of fundamental theories: Consider e.g. set theory with axiom of choice, foundations of logic, exact/full minimax for zero-sum games, quantum (field) theory, string theory, … Indeed, at least in physics, every new fundamental theory seems to
4 0.15218934 295 hunch net-2008-04-12-It Doesn’t Stop
Introduction: I’ve enjoyed the Terminator movies and show. Neglecting the whacky aspects (time travel and associated paradoxes), there is an enduring topic of discussion: how do people deal with intelligent machines (and vice versa)? In Terminator-land, the primary method for dealing with intelligent machines is to prevent them from being made. This approach works pretty badly, because a new angle on building an intelligent machine keeps coming up. This is partly a ploy for writer’s to avoid writing themselves out of a job, but there is a fundamental truth to it as well: preventing progress in research is hard. The United States, has been experimenting with trying to stop research on stem cells . It hasn’t worked very well—the net effect has been retarding research programs a bit, and exporting some research to other countries. Another less recent example was encryption technology, for which the United States generally did not encourage early public research and even discouraged as a mu
5 0.14900093 424 hunch net-2011-02-17-What does Watson mean?
Introduction: Watson convincingly beat the best champion Jeopardy! players. The apparent significance of this varies hugely, depending on your background knowledge about the related machine learning, NLP, and search technology. For a random person, this might seem evidence of serious machine intelligence, while for people working on the system itself, it probably seems like a reasonably good assemblage of existing technologies with several twists to make the entire system work. Above all, I think we should congratulate the people who managed to put together and execute this project—many years of effort by a diverse set of highly skilled people were needed to make this happen. In academia, it’s pretty difficult for one professor to assemble that quantity of talent, and in industry it’s rarely the case that such a capable group has both a worthwhile project and the support needed to pursue something like this for several years before success. Alina invited me to the Jeopardy watching party
6 0.1375078 286 hunch net-2008-01-25-Turing’s Club for Machine Learning
7 0.13315187 347 hunch net-2009-03-26-Machine Learning is too easy
8 0.11801822 332 hunch net-2008-12-23-Use of Learning Theory
9 0.11550077 235 hunch net-2007-03-03-All Models of Learning have Flaws
10 0.11363708 120 hunch net-2005-10-10-Predictive Search is Coming
11 0.11272525 440 hunch net-2011-08-06-Interesting thing at UAI 2011
12 0.11185073 250 hunch net-2007-06-23-Machine Learning Jobs are Growing on Trees
13 0.11178423 287 hunch net-2008-01-28-Sufficient Computation
14 0.10850773 132 hunch net-2005-11-26-The Design of an Optimal Research Environment
15 0.10722635 151 hunch net-2006-01-25-1 year
16 0.1044182 49 hunch net-2005-03-30-What can Type Theory teach us about Machine Learning?
17 0.10355983 370 hunch net-2009-09-18-Necessary and Sufficient Research
18 0.10298779 454 hunch net-2012-01-30-ICML Posters and Scope
19 0.098981395 435 hunch net-2011-05-16-Research Directions for Machine Learning and Algorithms
20 0.09853746 383 hunch net-2009-12-09-Inherent Uncertainty
topicId topicWeight
[(0, 0.279), (1, 0.033), (2, -0.08), (3, 0.09), (4, -0.013), (5, -0.006), (6, -0.024), (7, 0.028), (8, 0.035), (9, -0.072), (10, -0.088), (11, -0.057), (12, 0.036), (13, 0.053), (14, -0.035), (15, 0.041), (16, 0.135), (17, -0.09), (18, 0.091), (19, 0.084), (20, -0.009), (21, 0.107), (22, -0.129), (23, 0.164), (24, 0.096), (25, -0.066), (26, 0.026), (27, -0.135), (28, 0.077), (29, -0.001), (30, 0.083), (31, -0.038), (32, -0.041), (33, -0.069), (34, 0.122), (35, -0.006), (36, -0.098), (37, 0.019), (38, -0.013), (39, -0.022), (40, -0.071), (41, -0.054), (42, 0.056), (43, -0.132), (44, -0.059), (45, -0.011), (46, -0.006), (47, -0.005), (48, -0.007), (49, -0.052)]
simIndex simValue blogId blogTitle
same-blog 1 0.94519961 352 hunch net-2009-05-06-Machine Learning to AI
Introduction: I recently had fun discussions with both Vikash Mansinghka and Thomas Breuel about approaching AI with machine learning. The general interest in taking a crack at AI with machine learning seems to be rising on many fronts including DARPA . As a matter of history, there was a great deal of interest in AI which died down before I began research. There remain many projects and conferences spawned in this earlier AI wave, as well as a good bit of experience about what did not work, or at least did not work yet. Here are a few examples of failure modes that people seem to run into: Supply/Product confusion . Sometimes we think “Intelligences use X, so I’ll create X and have an Intelligence.” An example of this is the Cyc Project which inspires some people as “intelligences use ontologies, so I’ll create an ontology and a system using it to have an Intelligence.” The flaw here is that Intelligences create ontologies, which they use, and without the ability to create ont
2 0.93597454 380 hunch net-2009-11-29-AI Safety
Introduction: Dan Reeves introduced me to Michael Vassar who ran the Singularity Summit and educated me a bit on the subject of AI safety which the Singularity Institute has small grants for . I still believe that interstellar space travel is necessary for long term civilization survival, and the AI is necessary for interstellar space travel . On these grounds alone, we could judge that developing AI is much more safe than not. Nevertheless, there is a basic reasonable fear, as expressed by some commenters, that AI could go bad. A basic scenario starts with someone inventing an AI and telling it to make as much money as possible. The AI promptly starts trading in various markets to make money. To improve, it crafts a virus that takes over most of the world’s computers using it as a surveillance network so that it can always make the right decision. The AI also branches out into any form of distance work, taking over the entire outsourcing process for all jobs that are entirely di
3 0.91415983 353 hunch net-2009-05-08-Computability in Artificial Intelligence
Introduction: Normally I do not blog, but John kindly invited me to do so. Since computability issues play a major role in Artificial Intelligence and Machine Learning, I would like to take the opportunity to comment on that and raise some questions. The general attitude is that AI is about finding efficient smart algorithms. For large parts of machine learning, the same attitude is not too dangerous. If you want to concentrate on conceptual problems, simply become a statistician. There is no analogous escape for modern research on AI (as opposed to GOFAI rooted in logic). Let me show by analogy why limiting research to computational questions is bad for any field. Except in computer science, computational aspects play little role in the development of fundamental theories: Consider e.g. set theory with axiom of choice, foundations of logic, exact/full minimax for zero-sum games, quantum (field) theory, string theory, … Indeed, at least in physics, every new fundamental theory seems to
4 0.79432631 287 hunch net-2008-01-28-Sufficient Computation
Introduction: Do we have computer hardware sufficient for AI? This question is difficult to answer, but here’s a try: One way to achieve AI is by simulating a human brain. A human brain has about 10 15 synapses which operate at about 10 2 per second implying about 10 17 bit ops per second. A modern computer runs at 10 9 cycles/second and operates on 10 2 bits per cycle implying 10 11 bits processed per second. The gap here is only 6 orders of magnitude, which can be plausibly surpassed via cluster machines. For example, the BlueGene/L operates 10 5 nodes (one order of magnitude short). It’s peak recorded performance is about 0.5*10 15 FLOPS which translates to about 10 16 bit ops per second, which is nearly 10 17 . There are many criticisms (both positive and negative) for this argument. Simulation of a human brain might require substantially more detail. Perhaps an additional 10 2 is required per neuron. We may not need to simulate a human brain to achieve AI. Ther
5 0.75591731 295 hunch net-2008-04-12-It Doesn’t Stop
Introduction: I’ve enjoyed the Terminator movies and show. Neglecting the whacky aspects (time travel and associated paradoxes), there is an enduring topic of discussion: how do people deal with intelligent machines (and vice versa)? In Terminator-land, the primary method for dealing with intelligent machines is to prevent them from being made. This approach works pretty badly, because a new angle on building an intelligent machine keeps coming up. This is partly a ploy for writer’s to avoid writing themselves out of a job, but there is a fundamental truth to it as well: preventing progress in research is hard. The United States, has been experimenting with trying to stop research on stem cells . It hasn’t worked very well—the net effect has been retarding research programs a bit, and exporting some research to other countries. Another less recent example was encryption technology, for which the United States generally did not encourage early public research and even discouraged as a mu
6 0.72260338 153 hunch net-2006-02-02-Introspectionism as a Disease
7 0.65814614 424 hunch net-2011-02-17-What does Watson mean?
8 0.55360752 120 hunch net-2005-10-10-Predictive Search is Coming
9 0.54210436 493 hunch net-2014-02-16-Metacademy: a package manager for knowledge
10 0.54029948 168 hunch net-2006-04-02-Mad (Neuro)science
11 0.52590322 3 hunch net-2005-01-24-The Humanloop Spectrum of Machine Learning
12 0.49468765 328 hunch net-2008-11-26-Efficient Reinforcement Learning in MDPs
13 0.49035326 366 hunch net-2009-08-03-Carbon in Computer Science Research
14 0.48928723 49 hunch net-2005-03-30-What can Type Theory teach us about Machine Learning?
15 0.47646251 370 hunch net-2009-09-18-Necessary and Sufficient Research
16 0.46656439 286 hunch net-2008-01-25-Turing’s Club for Machine Learning
17 0.4644264 270 hunch net-2007-11-02-The Machine Learning Award goes to …
18 0.46014398 230 hunch net-2007-02-02-Thoughts regarding “Is machine learning different from statistics?”
19 0.45670372 250 hunch net-2007-06-23-Machine Learning Jobs are Growing on Trees
20 0.45426187 128 hunch net-2005-11-05-The design of a computing cluster
topicId topicWeight
[(10, 0.029), (27, 0.558), (38, 0.063), (53, 0.056), (55, 0.065), (56, 0.016), (58, 0.018), (64, 0.014), (66, 0.014), (94, 0.08), (95, 0.02)]
simIndex simValue blogId blogTitle
same-blog 1 0.99780732 352 hunch net-2009-05-06-Machine Learning to AI
Introduction: I recently had fun discussions with both Vikash Mansinghka and Thomas Breuel about approaching AI with machine learning. The general interest in taking a crack at AI with machine learning seems to be rising on many fronts including DARPA . As a matter of history, there was a great deal of interest in AI which died down before I began research. There remain many projects and conferences spawned in this earlier AI wave, as well as a good bit of experience about what did not work, or at least did not work yet. Here are a few examples of failure modes that people seem to run into: Supply/Product confusion . Sometimes we think “Intelligences use X, so I’ll create X and have an Intelligence.” An example of this is the Cyc Project which inspires some people as “intelligences use ontologies, so I’ll create an ontology and a system using it to have an Intelligence.” The flaw here is that Intelligences create ontologies, which they use, and without the ability to create ont
2 0.99080288 45 hunch net-2005-03-22-Active learning
Introduction: Often, unlabeled data is easy to come by but labels are expensive. For instance, if you’re building a speech recognizer, it’s easy enough to get raw speech samples — just walk around with a microphone — but labeling even one of these samples is a tedious process in which a human must examine the speech signal and carefully segment it into phonemes. In the field of active learning, the goal is as usual to construct an accurate classifier, but the labels of the data points are initially hidden and there is a charge for each label you want revealed. The hope is that by intelligent adaptive querying, you can get away with significantly fewer labels than you would need in a regular supervised learning framework. Here’s an example. Suppose the data lie on the real line, and the classifiers are simple thresholding functions, H = {h w }: h w (x) = 1 if x > w, and 0 otherwise. VC theory tells us that if the underlying distribution P can be classified perfectly by some hypothesis in H (
3 0.98917025 304 hunch net-2008-06-27-Reviewing Horror Stories
Introduction: Essentially everyone who writes research papers suffers rejections. They always sting immediately, but upon further reflection many of these rejections come to seem reasonable. Maybe the equations had too many typos or maybe the topic just isn’t as important as was originally thought. A few rejections do not come to seem acceptable, and these form the basis of reviewing horror stories, a great material for conversations. I’ve decided to share three of mine, now all safely a bit distant in the past. Prediction Theory for Classification Tutorial . This is a tutorial about tight sample complexity bounds for classification that I submitted to JMLR . The first decision I heard was a reject which appeared quite unjust to me—for example one of the reviewers appeared to claim that all the content was in standard statistics books. Upon further inquiry, several citations were given, none of which actually covered the content. Later, I was shocked to hear the paper was accepted. App
4 0.98839861 196 hunch net-2006-07-13-Regression vs. Classification as a Primitive
Introduction: For learning reductions we have been concentrating on reducing various complex learning problems to binary classification. This choice needs to be actively questioned, because it was not carefully considered. Binary clasification is learning a classifier c:X -> {0,1} so as to minimize the probability of being wrong, Pr x,y~D (c(x) y) . The primary alternative candidate seems to be squared error regression. In squared error regression, you learn a regressor s:X -> [0,1] so as to minimize squared error, E x,y~D (s(x)-y) 2 . It is difficult to judge one primitive against another. The judgement must at least partially be made on nontheoretical grounds because (essentially) we are evaluating a choice between two axioms/assumptions. These two primitives are significantly related. Classification can be reduced to regression in the obvious way: you use the regressor to predict D(y=1|x) , then threshold at 0.5 . For this simple reduction a squared error regret of r
5 0.98675793 400 hunch net-2010-06-13-The Good News on Exploration and Learning
Introduction: Consider the contextual bandit setting where, repeatedly: A context x is observed. An action a is taken given the context x . A reward r is observed, dependent on x and a . Where the goal of a learning agent is to find a policy for step 2 achieving a large expected reward. This setting is of obvious importance, because in the real world we typically make decisions based on some set of information and then get feedback only about the single action taken. It also fundamentally differs from supervised learning settings because knowing the value of one action is not equivalent to knowing the value of all actions. A decade ago the best machine learning techniques for this setting where implausibly inefficient. Dean Foster once told me he thought the area was a research sinkhole with little progress to be expected. Now we are on the verge of being able to routinely attack these problems, in almost exactly the same sense that we routinely attack bread and but
6 0.98282582 9 hunch net-2005-02-01-Watchword: Loss
7 0.98258293 308 hunch net-2008-07-06-To Dual or Not
8 0.98160034 172 hunch net-2006-04-14-JMLR is a success
9 0.98149848 288 hunch net-2008-02-10-Complexity Illness
10 0.97981179 341 hunch net-2009-02-04-Optimal Proxy Loss for Classification
11 0.97885835 245 hunch net-2007-05-12-Loss Function Semantics
12 0.97727156 247 hunch net-2007-06-14-Interesting Papers at COLT 2007
13 0.97532064 274 hunch net-2007-11-28-Computational Consequences of Classification
14 0.97185445 14 hunch net-2005-02-07-The State of the Reduction
15 0.97082543 293 hunch net-2008-03-23-Interactive Machine Learning
16 0.96953952 166 hunch net-2006-03-24-NLPers
17 0.96953952 246 hunch net-2007-06-13-Not Posting
18 0.96953952 418 hunch net-2010-12-02-Traffic Prediction Problem
19 0.96869963 244 hunch net-2007-05-09-The Missing Bound
20 0.96862555 311 hunch net-2008-07-26-Compositional Machine Learning Algorithm Design