Weakly-supervised analysis dictionary learning with cardinality constraints

You, Zeyu; Raich, Raviv; Fern, Xiaoli Z.; Kim, Jinsub

doi:10.1109/ssp.2016.7551850

Cited by 1 publication

(6 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In our preliminary work [16], a convolutive analysis dictionary learning approach under weak-supervision has been proposed. Here, we expand on the approach of [16]. We simplify the previous graphical model and re-derive all formulations.…”

Section: Discriminative Dictionary Learningmentioning

confidence: 99%

“…Due to the deterministic rule between (Y j−1 nt , N j−1 nt ) and (Y j n(2t−1) , N j n(2t−1) ), (Y j n(2t) , N j n(2t) ) as proposed in (15) and (16), P (Y j−1 nt , N j−1 nt |Y j n(2t−1) , N j n(2t−1) , Y j n(2t) , N j n(2t) ) = I(Y j−1 nt = j n(2t−1) ∪Y j n(2t) )I(N j−1 nt = N j n(2t−1) + N j n(2t) ). Combining the above the equations, we obtain the update rule of the forward message passing on tree as:…”

Section: Appendix G Derivation Of Forward Message Passing On Treementioning

confidence: 99%

“…We present the derivations in the new model in detail in this paper. Additionally, a novel tree reformulation of the graphical model (unavailable in [16]) is provided in this paper. Thirdly, due to the modeling variation, all derivations have been redone.…”

Section: B Analysis Dictionary Learningmentioning

confidence: 99%

“…This reformulation yields a near-linear exact probability calculation that alleviates the need for approximate inference. Our preliminary work on this topic was introduced in [16]. 1…”

Section: Introductionmentioning

confidence: 99%

“…The model in this paper provides a simplification of the model in[16] to make the new model elegant and clear. We present the derivations in the new model in detail in this paper.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Weakly Supervised Dictionary Learning

You

Raich

Fern

et al. 2018

IEEE Trans. Signal Process.

Self Cite

View full text Add to dashboard Cite

We present a probabilistic modeling and inference framework for discriminative analysis dictionary learning under a weak supervision setting. Dictionary learning approaches have been widely used for tasks such as low-level signal denoising and restoration as well as high-level classification tasks, which can be applied to audio and image analysis. Synthesis dictionary learning aims at jointly learning a dictionary and corresponding sparse coefficients to provide accurate data representation. This approach is useful for denoising and signal restoration, but may lead to sub-optimal classification performance. By contrast, analysis dictionary learning provides a transform that maps data to a sparse discriminative representation suitable for classification. We consider the problem of analysis dictionary learning for timeseries data under a weak supervision setting in which signals are assigned with a global label instead of an instantaneous label signal. We propose a discriminative probabilistic model that incorporates both label information and sparsity constraints on the underlying latent instantaneous label signal using cardinality control. We present the expectation maximization (EM) procedure for maximum likelihood estimation (MLE) of the proposed model. To facilitate a computationally efficient E-step, we propose both a chain and a novel tree graph reformulation of the graphical model. The performance of the proposed model is demonstrated on both synthetic and real-world data.

show abstract

Section: Discriminative Dictionary Learningmentioning

confidence: 99%

Section: Appendix G Derivation Of Forward Message Passing On Treementioning

confidence: 99%

Section: B Analysis Dictionary Learningmentioning

confidence: 99%

“…This reformulation yields a near-linear exact probability calculation that alleviates the need for approximate inference. Our preliminary work on this topic was introduced in [16]. 1…”

Section: Introductionmentioning

confidence: 99%

“…The model in this paper provides a simplification of the model in[16] to make the new model elegant and clear. We present the derivations in the new model in detail in this paper.…”

mentioning

confidence: 99%

See 3 more Smart Citations