WO2005111814A1 - A method of adaptive learning through pattern matching - Google Patents
A method of adaptive learning through pattern matching Download PDFInfo
- Publication number
- WO2005111814A1 WO2005111814A1 PCT/CN2004/000492 CN2004000492W WO2005111814A1 WO 2005111814 A1 WO2005111814 A1 WO 2005111814A1 CN 2004000492 W CN2004000492 W CN 2004000492W WO 2005111814 A1 WO2005111814 A1 WO 2005111814A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- events
- further characterized
- sequence
- event
- electrical
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
- G06N5/046—Forward inferencing; Production systems
- G06N5/047—Pattern matching networks; Rete networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Definitions
- a Method of Adaptive Learning through Pattern Matching This invention relates to a method of adaptive learning through pattern matching, which may advantageously be implemented in a data processing system, and a system adapted to work such a method.
- a premises to be managed by a system and a method according to the present invention may be wired up with a number of electrical and/or electronic sensors, detectors and devices, all of which being connected, monitored and controlled by a distributed network of controllers and servers communicating via a digital communications backbone, as detailed in Hong Kong Short-Term Patent No. HK 1052830.
- These devices, sensors and detectors may include, e.g. motion detectors, window/door contact switches, smoke detectors, light sensors, pressure sensors, sound sensors, etc. All events and occurrences (e.g. tripping of a motion detectors, turning on lights, walking past a corridor, etc.) in the premises are thus detectable by the system via these sensors and detectors and these events are constantly monitored by the system.
- Specific actions may be initiated and performed by the system upon the detection of an event, which are usually pre-defined or pre-scripted. It is not the purpose of this invention to depict how the actual detectors and sensors and monitoring equipment are to be connected, or how such detection is performed. Suffice to say that, the premises is assumed to have been installed with adequate detection and sensing equipment to enable rapid detection of any event happening in the premises by a management method and system according to the present invention.
- a method of operating a plurality of electrical and/or electronic devices connected with a data processing apparatus including the steps of (a) detecting occurrence of events of at least one of said devices; (b) recording data relating to at least some of said detected events in a database; (c) ordering said recorded events into at least one sequence of events; and (d) comparing a detected event with said sequence(s) of events.
- a system including a plurality of electrical and/or electronic devices connected with a data processing apparatus, including (a) means for detecting occurrence of events of at least one of said devices; (b) means for recording at least some of said detected events in a database; (c) means for ordering said recorded events into at least one sequence of events; and (d) means for comparing a detected event with said sequence(s) of events.
- Fig. 1 is a flow chart showing the procedure of event logging of a method and system according to an embodiment of the present invention.
- FIG. 2A and 2B are a flow chart showing the procedure of pattern matching and pattern revision of a method and system according to an embodiment of the present invention.
- the main purpose of the present invention is to introduce intelligence into, e.g. a system for the management of a premises, through a system allowing, and a method including, pattern matching and adaptive learning.
- the system upon detection of an event, the system can actually make intelligent decisions according to its own accumulated experience and learning without having to resort to pre-programmed sequences.
- a user enters the premises with his/her keycard.
- the system detects the scanning of the keycard at a card reader and notices the presence of the user.
- the system detects that the user turns on the lights in the room, sets the temperature control in the room to 21°C, and puts on a classical music album on the audio system.
- the premises management system records these events in the history log, and since no particular actions have been pre-programmed based on these events, the system will simply do nothing but continue to observe and record. Further assume that the user repeats this sequence of actions a few times, each time entering the premises with his/her keycard, turning on the same lights, setting the temperature control to 21°C, and putting on the same classical music album.
- the present intelligent management system will then infer the causal relationship (a "behavior" of the user) between entering the room with this user's keycard and his/her subsequent actions.
- the system can actually propose to turn on the lights, set climate control to 21°C and put on music without the user having to lift another finger.
- the system may automatically perform these actions for the user until the user specifically stops it, or the system may prompt the user for confirmation via speech messages announced on the speakers installed in the room, and obtain the user's confirmation via speech recognition via a microphone installed in the room.
- a system constantly monitors all the current state values of each connected device, and when the value of any state changes, it is automatically recorded as an event. At least some the following information will also be recorded with the event: time of event; identity of the device; - state values: previous and new; and any user identity associated with the event, e.g. keycard number mapped to a user's identity when the event is scanning of a keycard.
- the intelligent management system sorts all recorded events by time stamp order to create an event sequence. It then partitions the sequence based on geographical relevancy and yield sub-sequences that are more localized and more manageable.
- the premises consists of two rooms connected by a common corridor. Events generated by devices in one room are less relevant to events generated by devices in the other room, but events generated in the corridor are relevant to both rooms. This is because, in premises management, events that happen are typically localized geographically: when two events happen at the same time at two geographically diverse places, chances are much higher that two different users are responsible for mutually independent actions.
- the intelligent management system keeps a crosstab table with rows and columns, both consisting of different geographical zones in the premises.
- the table is filled with relevancy coefficients, between 0 and 1, with 0 meaning no relevancy at all, and 1 means total relevancy.
- Table 2 below shows exemplary relevancy coefficients of a premises with five zones, namely Zone A, Zone B, Zone C, Zone D and Zone E.
- Zone B will have a relevancy of 0.7 (high relevancy) to Zone C, but a device triggered in
- Zone C will have a relevancy of only 0.2 (low relevancy) to Zone B.
- the relevancy coefficients can be made asymmetric, so that more flexibility can be obtained when designing zonal structures.
- each device For each event recorded, its relevancy coefficient relative to each zone is multiplied by the magnitude of change in the state value for the event, as well as a pre-defined coefficient for that device (usually a number for magnitude scaling purposes) to arrive at the final relevancy index of this event for the zone in question. As such, each device should has its own coefficient.
- a temperature sensor yields the current temperature of the environment, and that the pre-defined coefficient for this sensor is 0.5, and its relevancy coefficient relative to Zone A is 0.7. Let us further assume that the temperature changes from 21°C to 25°C. The state value of this sensor will thus change from 21 to 25. The final relevancy index of this event for Zone A will thus be: Magnitude of change x Pre-defined coefficient x Relevancy coefficient
- the total event sequence is then sub-divided into sub-sequences, one for each zone in the premises.
- the events included in each zone's sub-sequence are those events with relevancy indices exceeding a pre-defined threshold value. Therefore, although there is only one event sub-sequence for each zone, any one event may be included in none, one, or more such sub-sequences, indicating that the event is of interest to nobody, to only one zone, or to more than one zone respectively.
- the threshold may be set to zero to allow all events to be relevant to all zones, regardless of where the event is recorded.
- the threshold for including this event in a particular zone's event subsequences is 1.0. In this case, this event will be included in the subsequences for Zones A, D and E, but not in those for Zones B or C. Of course, if the threshold is set to zero, then all events will be included in all zones. It is also the case that, in typical premises management scenarios, events happening over a long period of time are less relevant to each other, as compared with events happening within a short time from each other.
- a light being turned on now has no reasonable relevancy to another light being turned on one week later, but is very relevant to another light being turned on immediately after. This is because human behavior typically occurs in bursts of actions, one closely after another. For this reason, the management system will automatically discard events that are older than a pre-defined age (e.g. one week) in order to conserve computing and storage resources.
- This threshold age may be set to infinity, so that the system will always consider all events recorded in its database.
- a pattern is defined as a sequence of events that has a tendency to repeat, either in its entirety or in sequences that closely resemble the original pattern.
- the entire process of pattern recognition detects regularities in behavior, and these regularities enable the management system to "program itself to perform some regular activities on behalf of the user, as a result creating convenience and the perception of intelligence. Recognizing pattern in a time-based sequence is a well-studied computing problem.
- the current implementation makes use of a mixture of Dynamic Programming and Gibbs Sampling techniques for pattern matching purposes.
- the standard Gibbs Sampling algorithm is slightly modified to allow patterns of variable lengths, but with events happening within a pre-defined "time window.” This decision is based on the fact that human behavior occurs in bursts of activity; therefore, events occurring together within, say, 30 seconds or 1 minute, have much higher chances of being parts of a coherent whole than events happening 2 hours apart.
- the "time window” moves through the events sub-sequence for each zone instead of the standard-length sub-string in the standard Gibbs Sampling algorithm.
- Dynamic Programming is a heuristic search technique typically used for detecting missing information in sequences when matched against a pattern. It can also be used for detecting swaps and slight variations in ordering.
- the standard Gibbs Sampling comparison operation is replaced by a more involved Dynamic Programming operation so that event sequences that resemble each other and differ only slightly in ordering are counted as a successful match.
- the output of a Dynamic Programming operation includes a numeric score that enables the use of a pre-defined threshold to fine tune the pattern recognition algorithm. Setting a low threshold will enable more different patterns to be recognized, while setting a high threshold will yield fewer patterns, although such patterns resemble each other much more closely.
- An optional modification to the system is to only consider recognizing a pattern when the time of the event or the changes of state values of certain devices are within a pre-set range. For example, certain patterns are relevant only at certain times of day (e.g. turning on lights only when it is night time), or days of week (e.g. watering the plants only twice a week), or only when certain conditions are true (e.g. turning on lights only when it is dark).
- certain patterns are relevant only at certain times of day (e.g. turning on lights only when it is night time), or days of week (e.g. watering the plants only twice a week), or only when certain conditions are true (e.g. turning on lights only when it is dark).
- the Gibbs Sampling technique also yields a score for the pattern within the sequence.
- a threshold value may be used for rejecting patterns that are too far off from each other, and to include as valid only those patterns with scores that are above the threshold.
- pattern recognition always operates on the sub-sequence of each zone, the patterns recognized are always tied to a particular zone and are stored in the system's pattern database for future matching purposes.
- the intelligent management system will constantly monitor the incoming stream of detected events to see whether they match the beginning of one of the patterns stored in the database.
- any new event is screened through relevancy calculations, with irrelevant events, i.e. events with low relevancy indices, discarded.
- Dynamic Programming techniques are then employed to compare any new event with the beginning of corresponding patterns stored relative to the zone in question. The result of comparison is considered to be positive, and a pattern is thus to be matched, if its Dynamic Programming score exceeds a pre-defined threshold. With each pattern is also kept a respective success rate. If the beginning of a pattern is successfully matched with a new event, then the system may automatically perform all the remaining actions in the pattern or inform the user via speech generation for confirmation. Upon confirmation by the user, e.g. via speech recognition, the system will also update the success rate based on the response of the user. If the user confirms the system's suggestion, the success rate will be increased; whereas if the user rejects the system's suggestion, the success rate will be decreased.
- An important adaptive technique to enhance the learning capabilities of the system is to allow the system to dynamically alter the stored patterns based on actual occurrences. For example, suppose a pattern stored in the database is as follows:
- Zone John's bedroom Pattern: Keycard scanner detects John's keycard Unlock bedroom door Turn on bedroom lights Turn on audio system Put on song A
- the system will replace the particular action in the pattern with this variation.
- the system will rewrite the pattern as follows:
- Zone John's bedroom Pattern: Keycard scanner detects John's keycard Unlock bedroom door Turn on bedroom lights Turn on audio system Put on song B
- an adaptive learning, intelligent premise management system exhibits a perception of intelligence with its ability to recognize human behavior (i.e. patterns) and to offer to automate those activities that are repetitive. It also adapts to on-going runs by learning new variations and dynamically rewriting the database of recognized behavior.
- Figs. 1 and 2 The steps and procedure for the operation of a method and system according to the present invention is shown in the flow charts in Figs. 1 and 2.
- Fig. 1 when the system starts operation (Step 102), it will monitor the current states of all electronic and/or electrical devices connected to the system (Step 104), in particular to check if there has been any change in the value of the state of such devices (Step 106). Once a value change occurs, the event will be recorded or "logged" (Step 108), and its relevancy index for each zone in the system calculated (Step 110). The system will then check whether such calculated relevancy indices are above a pre-set threshold for each respective zone (Step 112).
- Step 114 the system will keep on monitoring the current states of all electronic and/or electrical devices connected to the system. If, on the other hand, there is any calculated relevancy index which is above the pre-set threshold for a particular zone, the event in question will be included in the sub-sequence of that particular zone (Step 114).
- Step 202 when the pattern matching function of the system starts (Step 202), which in fact operates simultaneously with the event logging and pattern recognition functions discussed above, the system will monitor all incoming detected events (Step 204). The system will check whether the calculated relevancy indices of such detected events are above a pre-set threshold (Step 206). Once a relevancy index of a detected event exceeds the pre-set threshold, the system will then determine whether the Dynamic Programming score of this event is above another pre-set threshold (Step 208). If not, the system will continue with the monitoring of incoming detected events (Step 204).
- Step 210 the system will proceed to perform all remaining actions in the matched pattern (Step 212).
- the system may be configured to seek confirmation from the user as to whether the remaining actions in the matched pattern are to be performed.
- the success rate will also be updated (Step 214).
- the system will then go on to check whether there is any change in the value of a device involved in the pattern within a pre-set time duration (Step 216). If not, the system will continue with the monitoring of incoming detected events (Step 204). If, on the other hand, the answer is in the affirmative, this event will be stored as a variant (Step 218). The system will then check whether this variant has occurred for up to a pre-set percentage, e.g. at least 80%, i.e. four times in out of the last five times when the pattern . was executed (Step 220). If not, the system will continue with the monitoring of incoming detected events (Step 204). If, on the other hand, the answer to this is in the affirmative, the particular action in the pattern will be replaced by the variant to generate a new sub-sequence, for future execution (Step 222).
- a pre-set percentage e.g. at least 80%, i.e. four times in out of the last five times when the pattern . was
- the current invention may be implemented in a vehicle of transport, e.g. an automobile, to give a perception that the vehicle has built-in "intelligence".
- a modern automobile contains a large number.of electronic and electrical devices, some for implementing certain functions (e.g. fuel injection control), some for monitoring (e.g. low-petrol sensor, speedometer), and some others for automation (e.g. window motors, seat position adjustment motors), etc.
- all such electronic and electrical devices are constantly monitored by a central data processing system to detect any and all state changes.
- a central data processing system to detect any and all state changes.
- RF Radio frequency
- this automobile has an RF based security key which, upon actuation, wirelessly transmits a security code to the RF receiver in the automobile.
- the security code is authenticated by the security system in the automobile, the door is unlocked and all these events will be logged in the adaptive learning system in the automobile.
- the driver upon entering the automobile, immediately adjusts the seat and rear view mirror positions to suit his physical built, turns on the radio and tunes it to Channel 2, and sets the climate control to 19°C. All such events (changes in the states of the electronic and/or electrical devices in the automobile) will, again, be logged by the system. Let us assume further that this same driver, subsequently, performs the same sequence of actions three more times.
- a master program is set to run continuously, its task being to perform pattern recognition based on logged event sequences, in the manner as discussed above. Let us assume that, after running a built-in Gibbs Sampling-based pattern extractor program, the pattern as shown in Table 4 below is found with a score exceeding the pre-defined minimum threshold:
- the system in the automobile may, after unlocking of the door, seek the driver's confirmation whether or not to continue to execute the remaining actions in the pattern. Once the driver's confirmation is received, the automobile will immediately adjust the seat position, rear view mirror position, turn on the radio, tune to the correct channel, and set the climate control, to the preferred setting and pattern. It can be seen that the only action that the Driver A has taken is unlocking of the door of the automobile by his security key, whereas the rest of the actions are all undertaken automatically by the system in the automobile, thus yielding the perception of "intelligence".
- a method according to the present invention may be implemented, and a system according to the present invention may be adopted, in the management of a computer network, e.g. to implement self-diagnosis and self-recovery functions, as illustrated below.
- a computer network is formed by "nodes" linked with one another into a topology.
- Each node represents a resource on the network, be it a computer (with data processing resources), a printer (with printing resources), a facsimile machine (with faxing resources), etc..
- Each node has a number of status variables, indicating its "health" and resource level. For this particular example, we will focus on the following devices in Table 5 below for illustration purposes:
- Computer A and Computer B may be of different types that cannot inter-operate
- Computer C may be of the same or interoperable type as Computer A, and since these three computers are connected with one another in the network, their geographical distances are irrelevant.
- Computer A and Computer C may be relevant to each other, whereas Computer B is relevant to neither of the other two computers.
- Table 6 below shows exemplary relevancy coefficients, in terms of the locations of the above devices: Table 6
Abstract
Description
Claims
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/CN2004/000492 WO2005111814A1 (en) | 2004-05-17 | 2004-05-17 | A method of adaptive learning through pattern matching |
US10/515,013 US7496545B2 (en) | 2004-05-17 | 2004-05-17 | Method of adaptive learning through pattern matching |
CN200480043049XA CN1954303B (en) | 2004-05-17 | 2004-05-17 | A method of adaptive learning through pattern matching |
HK07109199.3A HK1101290A1 (en) | 2004-05-17 | 2007-08-23 | A method of adaptive learning through pattern matching |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/CN2004/000492 WO2005111814A1 (en) | 2004-05-17 | 2004-05-17 | A method of adaptive learning through pattern matching |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2005111814A1 true WO2005111814A1 (en) | 2005-11-24 |
Family
ID=35394324
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2004/000492 WO2005111814A1 (en) | 2004-05-17 | 2004-05-17 | A method of adaptive learning through pattern matching |
Country Status (4)
Country | Link |
---|---|
US (1) | US7496545B2 (en) |
CN (1) | CN1954303B (en) |
HK (1) | HK1101290A1 (en) |
WO (1) | WO2005111814A1 (en) |
Families Citing this family (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7650282B1 (en) * | 2003-07-23 | 2010-01-19 | Nexidia Inc. | Word spotting score normalization |
US8700404B1 (en) | 2005-08-27 | 2014-04-15 | At&T Intellectual Property Ii, L.P. | System and method for using semantic and syntactic graphs for utterance classification |
US8752175B2 (en) * | 2008-10-31 | 2014-06-10 | Hewlett-Packard Development Company, L.P. | Method and apparatus for network intrusion detection |
US8010262B2 (en) * | 2009-10-21 | 2011-08-30 | Cnh America Llc | Apparatus and method for automatically controlling the settings of an adjustable crop residue spreader of an agricultural combine |
US8949179B2 (en) | 2012-04-23 | 2015-02-03 | Google, Inc. | Sharing and synchronizing electronically stored files |
US9239846B2 (en) | 2012-04-23 | 2016-01-19 | Google Inc. | Sharing and synchronizing electronically stored files |
US20130282830A1 (en) * | 2012-04-23 | 2013-10-24 | Google, Inc. | Sharing and synchronizing electronically stored files |
US9244934B2 (en) | 2012-04-23 | 2016-01-26 | Google Inc. | Sharing and synchronizing electronically stored files |
US9529818B2 (en) | 2012-04-23 | 2016-12-27 | Google Inc. | Sharing and synchronizing electronically stored files |
US20130346917A1 (en) * | 2012-06-22 | 2013-12-26 | Microsoft Corporation | Client application analytics |
US20140208214A1 (en) * | 2013-01-23 | 2014-07-24 | Gabriel D. Stern | Systems and methods for monitoring, visualizing, and managing physical devices and physical device locations |
WO2014169182A2 (en) | 2013-04-12 | 2014-10-16 | Dana Limited | Vehicle and operator guidance by pattern recognition |
US9509763B2 (en) | 2013-05-24 | 2016-11-29 | Qualcomm Incorporated | Delayed actions for a decentralized system of learning devices |
US9747554B2 (en) | 2013-05-24 | 2017-08-29 | Qualcomm Incorporated | Learning device with continuous configuration capability |
US9679491B2 (en) | 2013-05-24 | 2017-06-13 | Qualcomm Incorporated | Signaling device for teaching learning devices |
DE112014006409T5 (en) * | 2014-02-26 | 2016-12-08 | Mitsubishi Electric Corporation | Vehicle control device and vehicle control method |
US10154321B2 (en) * | 2014-05-26 | 2018-12-11 | Agt International Gmbh | System and method for registering sensors used in monitoring-systems |
CN105270296B (en) * | 2014-06-09 | 2017-11-28 | 源捷公司 | Vehicle learns interface |
US10298411B2 (en) | 2016-10-24 | 2019-05-21 | Crestron Electronics, Inc. | Building management system that determines building utilization |
US10389578B2 (en) * | 2017-03-06 | 2019-08-20 | International Business Machines Corporation | Learned response for alerts |
CN111190938B (en) * | 2019-12-26 | 2023-08-01 | 博彦科技股份有限公司 | Data analysis method, device, storage medium and processor |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5117992A (en) * | 1991-01-28 | 1992-06-02 | Virginia International Terminals, Inc. | System for learning control commands to robotically move a load, especially suitable for use in cranes to reduce load sway |
US6292729B2 (en) * | 1999-04-14 | 2001-09-18 | Deere & Company | Vehicle function management system |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO1997013186A1 (en) * | 1995-10-06 | 1997-04-10 | Southpower Ltd. | Device for the operation of appliances, utilities and services within a building |
US5809492A (en) * | 1996-04-09 | 1998-09-15 | At&T Corp. | Apparatus and method for defining rules for personal agents |
US6909921B1 (en) * | 2000-10-19 | 2005-06-21 | Destiny Networks, Inc. | Occupancy sensor and method for home automation system |
US6792319B1 (en) * | 2000-10-19 | 2004-09-14 | Destiny Networks, Inc. | Home automation system and method |
US6912429B1 (en) * | 2000-10-19 | 2005-06-28 | Destiny Networks, Inc. | Home automation system and method |
-
2004
- 2004-05-17 US US10/515,013 patent/US7496545B2/en not_active Expired - Fee Related
- 2004-05-17 CN CN200480043049XA patent/CN1954303B/en not_active Expired - Fee Related
- 2004-05-17 WO PCT/CN2004/000492 patent/WO2005111814A1/en active Application Filing
-
2007
- 2007-08-23 HK HK07109199.3A patent/HK1101290A1/en not_active IP Right Cessation
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5117992A (en) * | 1991-01-28 | 1992-06-02 | Virginia International Terminals, Inc. | System for learning control commands to robotically move a load, especially suitable for use in cranes to reduce load sway |
US6292729B2 (en) * | 1999-04-14 | 2001-09-18 | Deere & Company | Vehicle function management system |
Also Published As
Publication number | Publication date |
---|---|
US20060074666A1 (en) | 2006-04-06 |
CN1954303B (en) | 2010-05-05 |
HK1101290A1 (en) | 2007-10-12 |
CN1954303A (en) | 2007-04-25 |
US7496545B2 (en) | 2009-02-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US7496545B2 (en) | Method of adaptive learning through pattern matching | |
US11050577B2 (en) | Automatically learning and controlling connected devices | |
US7552030B2 (en) | System and method for learning patterns of behavior and operating a monitoring and response system based thereon | |
Alippi et al. | Just-in-time classifiers for recurrent concepts | |
Singh | Transfer of learning by composing solutions of elemental sequential tasks | |
TWI229787B (en) | System and method for automatic control device personalization | |
JP2003111157A (en) | Integrated controller, apparatus controlling method, and apparatus controlling program | |
Sklansky | Learning systems for automatic control | |
EP2609792B1 (en) | Automatic configuration of a lighting system | |
JPWO2007086475A1 (en) | Plant monitoring and control device | |
KR20090027000A (en) | Apparatus and method of constructing user behavior pattern based on the event log generated from the context aware system environment | |
EP2257912A2 (en) | Pattern classification system and method for collective learning | |
JPWO2020053953A1 (en) | Matching system and matching server | |
WO2017157438A1 (en) | Method and device for real-time network event processing | |
CN113164056A (en) | Sleep prediction method, device, storage medium and electronic equipment | |
JP7409080B2 (en) | Learning data generation method, learning data generation program, and information processing device | |
Manna et al. | Learning occupancy in single person offices with mixtures of multi-lag markov chains | |
WO2020168444A1 (en) | Sleep prediction method and apparatus, storage medium, and electronic device | |
Mozer et al. | Parsing the stream of time: The value of event-based segmentation in a complex real-world control problem | |
CN112061908B (en) | Elevator control method and system | |
US20040107173A1 (en) | Operating system using artificial intelligence processing | |
WO2005018129A2 (en) | Improved gesture recognition for pointing devices | |
CN110874875B (en) | Door lock control method and device | |
CN113113008A (en) | Method for controlling fan through voice | |
US20210152579A1 (en) | Building security analysis system with site-independent signature generation for predictive security analysis |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 200480043049.X Country of ref document: CN |
|
ENP | Entry into the national phase |
Ref document number: 2006074666 Country of ref document: US Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 10515013 Country of ref document: US |
|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BW BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE EG ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NA NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL SY TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): GM KE LS MW MZ NA SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IT LU MC NL PL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
DPEN | Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed from 20040101) | ||
WWP | Wipo information: published in national office |
Ref document number: 10515013 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWW | Wipo information: withdrawn in national office |
Ref document number: DE |
|
122 | Ep: pct application non-entry in european phase |