Krzysztof Czarnowski - Academia.edu (original) (raw)
Related Authors
Graduate Center of the City University of New York
Uploads
Papers by Krzysztof Czarnowski
Demonstratio Mathematica, 1997
This paper deals with characterization of sets of solutions of equations in locally convex linear... more This paper deals with characterization of sets of solutions of equations in locally convex linear topological spaces, or, to be more specific, in Bo spaces. We use topological degree methods to obtain our main Theorem (16). Theorem (16) is a generalization of Theorem [4;(2.2)], which applies to fixed point sets of compact maps in Bo spaces, to a broader class of "k-setcontractive" maps. It goes parallel to a theorem of W. V. Petryshyn [11] on fixed point sets properties of some k-set-contractions in Banach spaces. The required extension of the Banach space notion of measure of noncompactness and k-set-contraction to the case of a Bo space is done in the first part of the paper. Next we use the ideas of R. D. Nussbaum [10] to define a topological degree for some k-set-contractions in Bo spaces. The theory of k-set-contractive and more general condensing maps in locally convex linear topological spaces is also given in the paper of B. N. Sadovskii [12]. It should be stated t...
Annales Polonici Mathematici, 1994
Interspeech 2019
This paper describes an approach for intent classification and tagging on embedded devices, such ... more This paper describes an approach for intent classification and tagging on embedded devices, such as smart watches. We describe a technique to train neuronal networks where the final neuronal network weights are binary. This enables memory bandwidth optimized inference and efficient computation even on constrained/embedded platforms. The flow of the approach is as follows: tf-idf word selection method reduces the number of overall weights. Bag-of-Words features are used with a feedforward and recurrent neuronal network for intent classification and tagging, respectively. A novel double Gaussian based regularization term is used to train the network. Finally, the weights are almost clipped lossless to −1 or 1 which results in a tiny binary neuronal network for intent classification and tagging. Our technique is evaluated using a text corpus of transcribed and annotated voice queries. The test domain is "lights control". We compare the intent and tagging accuracy of the ultra-compact binary neuronal network with our baseline system. The novel approach yields comparable accuracy but reduces the model size by a factor of 16: from 160kB to 10kB.
Nonlinear Analysis: Theory, Methods & Applications, 1996
Nonlinear Analysis: Theory, Methods & Applications, 1996
Journal of Mathematical Analysis and Applications, 1991
Demonstratio Mathematica, 1997
This paper deals with characterization of sets of solutions of equations in locally convex linear... more This paper deals with characterization of sets of solutions of equations in locally convex linear topological spaces, or, to be more specific, in Bo spaces. We use topological degree methods to obtain our main Theorem (16). Theorem (16) is a generalization of Theorem [4;(2.2)], which applies to fixed point sets of compact maps in Bo spaces, to a broader class of "k-setcontractive" maps. It goes parallel to a theorem of W. V. Petryshyn [11] on fixed point sets properties of some k-set-contractions in Banach spaces. The required extension of the Banach space notion of measure of noncompactness and k-set-contraction to the case of a Bo space is done in the first part of the paper. Next we use the ideas of R. D. Nussbaum [10] to define a topological degree for some k-set-contractions in Bo spaces. The theory of k-set-contractive and more general condensing maps in locally convex linear topological spaces is also given in the paper of B. N. Sadovskii [12]. It should be stated t...
Annales Polonici Mathematici, 1994
Interspeech 2019
This paper describes an approach for intent classification and tagging on embedded devices, such ... more This paper describes an approach for intent classification and tagging on embedded devices, such as smart watches. We describe a technique to train neuronal networks where the final neuronal network weights are binary. This enables memory bandwidth optimized inference and efficient computation even on constrained/embedded platforms. The flow of the approach is as follows: tf-idf word selection method reduces the number of overall weights. Bag-of-Words features are used with a feedforward and recurrent neuronal network for intent classification and tagging, respectively. A novel double Gaussian based regularization term is used to train the network. Finally, the weights are almost clipped lossless to −1 or 1 which results in a tiny binary neuronal network for intent classification and tagging. Our technique is evaluated using a text corpus of transcribed and annotated voice queries. The test domain is "lights control". We compare the intent and tagging accuracy of the ultra-compact binary neuronal network with our baseline system. The novel approach yields comparable accuracy but reduces the model size by a factor of 16: from 160kB to 10kB.
Nonlinear Analysis: Theory, Methods & Applications, 1996
Nonlinear Analysis: Theory, Methods & Applications, 1996
Journal of Mathematical Analysis and Applications, 1991