Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

Co-Planar Stereotaxic Atlas of the Human Brain—3-Dimensional Proportional System: An Approach to Cerebral Imaging, J. Talairach, P. Tournoux. Georg Thieme Verlag, New York (1988), 122 pp., 130 figs. DM 268

Nobel Lecture: Prions

Sleep improves the consolidation of both declarative and non-declarative memories. Diekelmann and Born discuss the potential mechanisms through which slow wave sleep and rapid eye movement sleep support system and synaptic consolidation. Sleep has been identified as a state that optimizes the consolidation of newly acquired information in memory, depending on the specific conditions of learning and the timing of sleep. Consolidation during sleep promotes both quantitative and qualitative changes of memory representations. Through specific patterns of neuromodulatory activity and electric field potential oscillations, slow-wave sleep (SWS) and rapid eye movement (REM) sleep support system consolidation and synaptic consolidation, respectively. During SWS, slow oscillations, spindles and ripples — at minimum cholinergic activity — coordinate the re-activation and redistribution of hippocampus-dependent memories to neocortical sites, whereas during REM sleep, local increases in plasticity-related immediate-early gene activity — at high cholinergic and theta activity — might favour the subsequent synaptic consolidation of memories in the cortex.

The memory function of sleep

OBJECTIVES: To inform the debate over whether human sleep can be chronically reduced without consequences, we conducted a dose-response chronic sleep restriction experiment in which waking neurobehavioral and sleep physiological functions were monitored and compared to those for total sleep deprivation. DESIGN: The chronic sleep restriction experiment involved randomization to one of three sleep doses (4 h, 6 h, or 8 h time in bed per night), which were maintained for 14 consecutive days. The total sleep deprivation experiment involved 3 nights without sleep (0 h time in bed). Each study also involved 3 baseline (pre-deprivation) days and 3 recovery days. SETTING: Both experiments were conducted under standardized laboratory conditions with continuous behavioral, physiological and medical monitoring. PARTICIPANTS: A total of n = 48 healthy adults (ages 21-38) participated in the experiments. INTERVENTIONS: Noctumal sleep periods were restricted to 8 h, 6 h or 4 h per day for 14 days, or to 0 h for 3 days. All other sleep was prohibited. RESULTS: Chronic restriction of sleep periods to 4 h or 6 h per night over 14 consecutive days resulted in significant cumulative, dose-dependent deficits in cognitive performance on all tasks. Subjective sleepiness ratings showed an acute response to sleep restriction but only small further increases on subsequent days, and did not significantly differentiate the 6 h and 4 h conditions. Polysomnographic variables and delta power in the non-REM sleep EEG-a putative marker of sleep homeostasis--displayed an acute response to sleep restriction with negligible further changes across the 14 restricted nights. Comparison of chronic sleep restriction to total sleep deprivation showed that the latter resulted in disproportionately large waking neurobehavioral and sleep delta power responses relative to how much sleep was lost. A statistical model revealed that, regardless of the mode of sleep deprivation, lapses in behavioral alertness were near-linearly related to the cumulative duration of wakefulness in excess of 15.84 h (s.e. 0.73 h). CONCLUSIONS: Since chronic restriction of sleep to 6 h or less per night produced cognitive performance deficits equivalent to up to 2 nights of total sleep deprivation, it appears that even relatively moderate sleep restriction can seriously impair waking neurobehavioral functions in healthy adults. Sleepiness ratings suggest that subjects were largely unaware of these increasing cognitive deficits, which may explain why the impact of chronic sleep restriction on waking cognitive functions is often assumed to be benign. Physiological sleep responses to chronic restriction did not mirror waking neurobehavioral responses, but cumulative wakefulness in excess of a 15.84 h predicted performance lapses across all four experimental conditions. This suggests that sleep debt is perhaps best understood as resulting in additional wakefulness that has a neurobiological "cost" which accumulates over time.

/pdf/the-cumulative-cost-of-additional-wakefulness-dose-response-3yfrcni48x.pdf

The Cumulative Cost of Additional Wakefulness: Dose-Response Effects on Neurobehavioral Functions and Sleep Physiology From Chronic Sleep Restriction and Total Sleep Deprivation

According to the two-process model of sleep regulation, the timing and structure of sleep are determined by the interaction of a homeostatic and a circadian process. The original qualitative model was elaborated to quantitative versions that included the ultradian dynamics of sleep in relation to the non-REM-REM sleep cycle. The time course of EEG slow-wave activity, the major marker of non-REM sleep homeostasis, as well as daytime alertness were simulated successfully for a considerable number of experimental protocols. They include sleep after partial sleep deprivation and daytime napping, sleep in habitual short and long sleepers, and alertness in a forced desynchrony protocol or during an extended photoperiod. Simulations revealed that internal desynchronization can be obtained for different shapes of the thresholds. New developments include the analysis of the waking EEG to delineate homeostatic and circadian processes, studies of REM sleep homeostasis, and recent evidence for local, use-dependent sleep processes. Moreover, nonlinear interactions between homeostatic and circadian processes were identified. In the past two decades, models have contributed considerably to conceptualizing and analyzing the major processes underlying sleep regulation, and they are likely to play an important role in future advances in the field.

/pdf/sleep-homeostasis-and-models-of-sleep-regulation-2kjl8u7e7t.pdf

Sleep homeostasis and models of sleep regulation

THERE is a wealth of data supporting a central role for the prion protein (PrP) in the neurodegenerative prion diseases of both humans and other species1, yet the normal function of PrP, which is expressed at the cell surface of neurons and glial cells2,3, is unknown. It has been speculated that neuropathology may be due to loss of normal function of PrP (ref. 4). Here we show that in mice devoid of PrP there is an alteration in both circadian activity rhythms and sleep patterns. To our knowledge, this is the first null mutation that has been shown to affect sleep regulation and our results indicate that the pathology of at least one of the inherited prion diseases, fatal familial insomnia5, where there is a profound alteration in sleep and the daily rhythms of many hormones6–10, may be related to the normal function of the prion protein.

Peter Achermann

Papers

Sleep homeostasis and models of sleep regulation

Altered circadian activity rhythms and sleep in mice devoid of prion protein.

Low-frequency (< 1 Hz) oscillations in the human sleep electroencephalogram.

A model of human sleep homeostasis based on eeg slow-wave activity - quantitative comparison of data and simulations

Dual electroencephalogram markers of human sleep homeostasis: correlation between theta activity in waking and slow-wave activity in sleep.