you uh i'm technical technique you in a from the uh yeah i T in japan that are are can can say shame but to feature extraction for three major jocks or something in in this walk we focus on the sense of device or or or three mean jokes or meter i don't such X or a is becoming a a a are sensitive by since it is and very in more so a smart phone such as i phones a i mean you of you might have such yeah smart phones i think and and uh and the same provocations use in there are such a signal meter is a screen rotation by detecting the orientation all utterances of the by source model for uh like these um this is a very convenient and useful for users browsing but uh i saying that beyond size seen propagation the three main jocks zero meter in smart phones has a potential to provide more source so katie pairs application or directly to the human action for examples uh action recognition for identifying data sparseness situations on that path somewhat identification such as by kate uh for security and so on i don't by in case oh at recognising human options the problem is that there are three dimensional acceleration signals are rotated according to the device orientation oh falls are a simple obligations and such a screen or stations it is forever used but uh for recognizing that human actions it is a big problem he calls the human options are you don't want to festivals orientation so in this walk or or uh our core is to propose an efficient method for extracting rotation invariant features for three-dimensional acceleration signals without any preprocessing for recognizing human actions independently of the device orientation uh you know charlie T uh the rent that the bias is device or an additional is changed uh there are also acceleration signals also rotate it even for the same option so in this case and and tsar rotation invariance means that we can't ten the same features from the same options no what the how that device or oriented the proposed missile is based on that frequency characteristics proof do used by it's are fourier transform we can see there are the corporation a long all frequencies that not sure tend to a or doing prospects on how right oh of us let me uh a views the to rate to walks fast ms or is based on the a simple statistics features these so a fast is to me is uh a vector or you know to the by the uh by applying the moving average to the of south a a and signal and the and there uh uh four kinds of statistics uh such as a mean and uh a standard deviations and uh are mean on mark's not a power created from three types of and uh a feature body uh such as uh it's normal of the observation big so and for that um course that between the oh so some back to a on the estimated graph depicted Z and uh uh second missile is based on the way it's features this look or so of uh how fast is uh up to the high so a C eight to zero so a source some signals um then uh a us at the level all way break a car create from the from those uh principal at three priest can point in these two missiles and uh the start is got preprocessing a you know these by a a bit strings uh and probably in a to compensate so and as you know station uh but such pre-processing required that and trying the of a signal data for statistical a stable words that so in this work we propose a novel feature extraction of some um sort which is in can three wrote at embedded and in the proposed missile so we focus on the frequency you relations which is uh useful to effectively characterise a few more options be calls the us assume a options are mainly consisting of a a cyclic a P all the motion such as a walking running on so on so two weeks of the frequency motion we have lies is that if if T actually short on if is T to to their of the of the house oh are some signal and uh we use uh for so you know two describes the proposed missile is do you know that are of obtained three main jocks or some signal in uh X Y and Z axes if if all me a in the case or so we the main general fourier if okay frequency see features also a a C all make and the in this case it up to a is a fruit of T two there and as read them engines uh in uh i X Y and the respective not and a joint three so was a fine oh uh long he's uh that's F of which in the case of free feature metrics of so a T is three D a frequency features F oh this rights shows a a core of the proposed missile still and uh we can is the correlation metrics of as the fourier features F which is a given by this a that that this four and uh and uh what don't here shows the process to construct a correlation metrics are which is and by and semantic matt six and here uh starting in also complex conjugate kate sports so in this course three and that time no am and a corresponding to the old you know you power to because the corporations between the same identical or or a frequency features as resulting in uh square and absolute areas on the other hand uh the uh of time diagonal elements uh cross-correlations between various kinds little frequency features so we can say that this improve call matt three not it contains not only is a all the emperor spectrum but also the uh cross corporation and which trained betas as frequencies so the proposed feature are are sort the body is all of a about trying to sorry right hand good components of the same matt are a uh i as you can see this is very simple formulation and the you can easy improvement it so uh uh the problem a a computational cost is also a little next uh we show was the invariance properties is pro involved that proposed missile there are two kinds of in there S uh fast one is the invariance to rotation that is the local main can someone in this uh in this work and uh and you rotation mark X is denoted by a uh uh as a a and not so that uh this or the semantic sees a a you only needed and uh by using this for the semantics a and is signal on the can so you can't or digits and three features that are describes so that D so uh the correlation metrics are all was uh rotate is signal now a is identical to so all of the or or no not to rotate data signal um by using these you type deals that also i rotation matrix so propose a feature is uh rotation in there next uh a second one is the in to the a temporal six as that is then try that in this case a we you know so then pride be eight by using the top and by using this time uh and there are a right signal on the can see can do that a features that are describe a that these so the of a select bar use all of the corporation board and a point as as that is little or fee propose features is also identical to that all the all reason not do rate signal so our proposed feature is also in there and to ten proceed uh such ten perceive is a a close by R short down if T so this and a in there as two ten it is uh a a useful in prague finally and we show so that we need to reduce there a post processing in do the proposed missile so of that of the original signals uh uh actually role and transpose by it is uh transfer function do to bear as min here such a causing which is depending on how to are the pisces score and yeah action eight is a action is not a directory of self but some who estimate estimated and such a scroll thing a if was transfer function is denoted by B and there's the of fourier of features of the uh uh of sub two signals is also trust full by using this at transfer phone function B that the and so uh a therefore for a we so a little all as a and propose a features to two in you you a price a fixed codes by the transfer function be the little little of the proposed feature is decomposed into these you is four in these phones uh the effects of the transfer function is its represent two that this bias and this is the same much patient not seen the some features full audio processing therefore are the in your must made on is such as that feature this unit that now is could start this guy can out that those by is so uh this is their experiments that were conducted in this walk and of the task or use a to identify a few more or human gait with that any control for this experiment we you only ask the uh fifty eight are sounds to three you walk in daily life a with holding i've phones in their hearts so we we this C is uh are quite challenging task for i in to find that a few more the experiment or set there please uh uh that's for roles we have to lies are seventy frames running windows for short time we for T with five T five frames tape side and uh and uh we and the except these as from there lining windows are classified into fifty eight to much cross um a by using a uh a nearest the means in their space of which are this content analysis i'm the oh is that i C is classified by using the mustard to boarding and uh we miss as the accuracy averaged a course the power phones by using C for course patient yeah a fast we shows that performance is an the setting sort of the proposed missile it's of uh three feature has are decomposed into two D C and that is can point uh that these you component is a zero frequency features the is can see that to be that gravity T pick so and so and these for kinds of was sitting that can see uh a C components only a and uh is C plus dc components which is a focal point and uh is C components with a little post processing and uh a C press D second points with a a little cost uh a processing so uh for the uh that are not feature as an which corresponds to that they re prospects and um by taking a low and is there a performance is and if the entry increased and that the best performance is of ten of and uh uh a a a a C of uh K C with a little sitting and the next one is uh a uh of that no features and in this case the best performance is of ten in the second or and a is the press C component but uh in for these an L of that in a few she shows by taking a little can and ends up performance is like three rate it and and up and a reason is that to uh roughly speaking and by taking a little the corporations and B to and there's frequency features P its prey got broken in to are just summation and uh this and the power or is also degraded we think so and finally i as a for the for feature and uh average of D two or is out and the best performance it's of tend well uh in the fading all its C plus T with a a a a little post processing uh you on show and that we also can cut is uh that a novel features on of that in features in the best sitting the performance is a for are improved and next we shows that a performance comparison to does not to missile and uh uh a as you can see the proposed missile is a significant three or to the other two minutes and these two other men use Z as a preprocessing four uh normalising norm are there is some signals orientation but uh our proposed missile and to not use that's uh uh uh uh a preprocessing and present also the proposed minutes so is uh or so competition a uh a sorry uh the proposed missile and or so yeah a the proposed a have a complete that cost in the proposed missile is also a little compared to these to to other missile and uh this is that something that supplement the experiment results uh in there some we did pay paper we only up to the classifier parser this comment comment ice is but in this case we also lies a go as far as of linear know it's fee and to the proposed minutes so propose a feature the performance is almost the same as that of the uh and this kind of analysis based cost it only the a zero point one percent is increased so what the conclusion uh we have proposed a mess so to so the road ensuring bad features from three to means know separation seen the proposed missile is based on the uh correlation metrics of forty if we can see feature containing a that prospects so in the that when no elements and the cross correlations in of that and i am and and the proposed feature is in parody invariant to rotation i there is a and perceived result any as the whole pretty uh preprocessing and in the experiments only if you might be shot yeah identifications sort of proposed so these superior to the other two missile and uh and there uh it should be not use that uh the proposed minutes so is a general so so it is a go to a break up to to the other tasks using their are there is some signal such as that a segment being on the recognise in there as human option and this is our feature walks i thank you for a