Tag Archives: mir

A different white-box deep architecture? Is it possible?

I recently published a paper more thoroughly explaining what I have been developing as a part of my PhD research at the Laboratory for Computer Graphics and Multimedia, Faculty of Computer and Information Science.

The paper describes an alternative deep architecture. Those of you who are involved in recent machine learning research probably know neural-network-based deep architectures, which recently gained a lot of attention from general public due to their results in a variety of fields. However, they are not perfect (at least I think so … :) ). To put it in a simple description, these networks lack transparence. We know how they work, but cannot see what they learn and thus often use them as black boxes (provide input, get output, don’t care about the process). We propose a different kind of deep architecture based on compositionality (large complex things are built of small simple things).

We tested our model on music information retrieval tasks and are currently seeking problems in other scientific fields dealing with big data – please message me for suggestions or potential collaboration.

If you are interested, I welcome you to read our paper, (freely) available here: http://dx.doi.org/10.1371/journal.pone.0169411

For a quick overview, I attached the paper’s abstract below:

Robust Real-Time Music Transcription with a Compositional Hierarchical Model

The paper presents a new compositional hierarchical model for robust music transcription. Its main features are unsupervised learning of a hierarchical representation of input data, transparency, which enables insights into the learned representation, as well as robustness and speed which make it suitable for real-world and real-time use.

The model consists of multiple layers, each composed of a number of parts. The hierarchical nature of the model corresponds well to hierarchical structures in music. The parts in lower layers correspond to low-level concepts (e.g. tone partials), while the parts in higher layers combine lower-level representations into more complex concepts (tones, chords). The layers are learned in an unsupervised manner from music signals. Parts in each layer are compositions of parts from previous layers based on statistical co-occurrences as the driving force of the learning process.

In the paper, we present the model’s structure and compare it to other hierarchical approaches in the field of music information retrieval. We evaluate the model’s performance for the multiple fundamental frequency estimation. Finally, we elaborate on extensions of the model towards other music information retrieval tasks.

Full paper is available here: http://dx.doi.org/10.1371/journal.pone.0169411

Češka – day 5 in 6

Včeraj je bil še en dan konference. Veliko predavanj, ki porodijo nove ideje. Zvečer smo šli na pivo z Nemci in IJSjevci (research purposes :) ).

5 times 2-fold cross validation oz. beer tasting

5 times 2-fold cross validation oz. beer tasting

Čehi so čudni. Večina kulture je tuje, tudi suvenirji. So pa mahnjeni na zlatnino.

Povsod je vse vsaj 50% "znižano".

Povsod je vse vsaj 50% “znižano”.

Tipičen čeh igra na dude

Češka narodna noša

Češka narodna noša

Izredno negujejo kitajsko kulturo neuporabnih stvari, ki jih veselo prodajajo :)

Izredno negujejo kitajsko kulturo neuporabnih stvari, ki jih veselo prodajajo :)

Nikakor pa ne pozabijo na zgodovino

Nikakor pa ne pozabijo na zgodovino

Nekaj čudnega je v zraku

Nekaj čudnega je v zraku

Danes je zadnji dan predavanj. Nekaj zanimivih utrinkov …

Ker je vse "deep", se tudi zagovorniki drugih pristopov trudijo utrditi svoj položaj.

Ker je vse “deep”, se tudi zagovorniki drugih pristopov trudijo utrditi svoj položaj.

Še en Machine Learning and Music "workshop". 2 prispevka v enem workshopu od iste avtorice. Jaz sem bil edini iz MIR scene na tej delavnici.

Še en Machine Learning and Music “workshop”. 2 prispevka v enem workshopu od iste avtorice. Jaz sem bil edini iz MIR scene na tej delavnici.

 

Češka – day 4

Čez dan konferenca, zvečer pa konferenčno srečanje. Par sličic konference, več pa Prage.

Praga

Praga

En izmed mnogih stolpov.

En izmed mnogih stolpov.

Azijci. En izmed bolj razumljivih slajdov.

Azijci. En izmed bolj razumljivih slajdov.

Keynote speaker. Naslov: "using machine learning powers for good". Delo: kako izmolzti več denarja iz volilcev za financiranje predsedniške kampanije. Evropejci imamo drugačno percepcijo dobrega.

Keynote speaker. Naslov: “using machine learning powers for good”. Delo: kako izmolzti več denarja iz volilcev za financiranje predsedniške kampanije. Evropejci imamo drugačno percepcijo dobrega.

 

Na vrhu gradu te čaka starbucks. Čehi so večino svojih dobrih lokacij prodali tujcem.

Na vrhu gradu te čaka starbucks. Čehi so večino svojih dobrih lokacij prodali tujcem.

Ljubljanica^2

Ljubljanica^2

Huda ura za v dnevno sobo. Ko posveti sonce, pokaže senca na astrološko znamenje. Poznavalci na 3883 nato izračunajo tarifo klicev njihovih specialistov za očiščenje čaker in nosnih por.

Huda ura za v dnevno sobo. Ko posveti sonce, pokaže senca na astrološko znamenje. Poznavalci na 3883 nato izračunajo tarifo klicev njihovih specialistov za očiščenje čaker in nosnih por.

Danes sem šel med odmorom za kosilo proti centru in si ogledal suvenirčke. Na žalost res (razen beherovke) ni kaj pametnega za kupit (saj lahko večino naročim preko dealextrema). Je pa veliko nakita (predvsem steklenega – ista serija kot tisti nakit na poletnih obmorskih štantih v Istrskih mestecih).

Po popoldanskem sessionu bom napadel manj znane lokacije, kjer naj bi se dalo dobit bolj pristne stvari Češke in ne “Češke” stvari.