-
Unlocking Insights into Business Trajectories with Transformer-based Spatio-temporal Data Analysis
Authors:
Muhammad Arslan,
Christophe Cruz
Abstract:
The world of business is constantly evolving and staying ahead of the curve requires a deep understanding of market trends and performance. This article addresses this requirement by modeling business trajectories using news articles data.
The world of business is constantly evolving and staying ahead of the curve requires a deep understanding of market trends and performance. This article addresses this requirement by modeling business trajectories using news articles data.
△ Less
Submitted 7 June, 2023;
originally announced June 2023.
-
Imbalanced Multi-label Classification for Business-related Text with Moderately Large Label Spaces
Authors:
Muhammad Arslan,
Christophe Cruz
Abstract:
In this study, we compared the performance of four different methods for multi label text classification using a specific imbalanced business dataset. The four methods we evaluated were fine tuned BERT, Binary Relevance, Classifier Chains, and Label Powerset. The results show that fine tuned BERT outperforms the other three methods by a significant margin, achieving high values of accuracy, F1 Sco…
▽ More
In this study, we compared the performance of four different methods for multi label text classification using a specific imbalanced business dataset. The four methods we evaluated were fine tuned BERT, Binary Relevance, Classifier Chains, and Label Powerset. The results show that fine tuned BERT outperforms the other three methods by a significant margin, achieving high values of accuracy, F1 Score, Precision, and Recall. Binary Relevance also performs well on this dataset, while Classifier Chains and Label Powerset demonstrate relatively poor performance. These findings highlight the effectiveness of fine tuned BERT for multi label text classification tasks, and suggest that it may be a useful tool for businesses seeking to analyze complex and multifaceted texts.
△ Less
Submitted 12 June, 2023;
originally announced June 2023.
-
Complexity of Shapes Embedded in ${\mathbb Z^n}$ with a Bias Towards Squares
Authors:
M. Ferhat Arslan,
Sibel Tari
Abstract:
Shape complexity is a hard-to-quantify quality, mainly due to its relative nature. Biased by Euclidean thinking, circles are commonly considered as the simplest. However, their constructions as digital images are only approximations to the ideal form. Consequently, complexity orders computed in reference to circle are unstable. Unlike circles which lose their circleness in digital images, squares…
▽ More
Shape complexity is a hard-to-quantify quality, mainly due to its relative nature. Biased by Euclidean thinking, circles are commonly considered as the simplest. However, their constructions as digital images are only approximations to the ideal form. Consequently, complexity orders computed in reference to circle are unstable. Unlike circles which lose their circleness in digital images, squares retain their qualities. Hence, we consider squares (hypercubes in $\mathbb Z^n$) to be the simplest shapes relative to which complexity orders are constructed. Using the connection between $L^\infty$ norm and squares we effectively encode squareness-adapted simplification through which we obtain multi-scale complexity measure, where scale determines the level of interest to the boundary. The emergent scale above which the effect of a boundary feature (appendage) disappears is related to the ratio of the contacting width of the appendage to that of the main body. We discuss what zero complexity implies in terms of information repetition and constructibility and what kind of shapes in addition to squares have zero complexity.
△ Less
Submitted 16 March, 2020;
originally announced March 2020.
-
Optimized Sharing of Coefficients in Parallel Filter Banks
Authors:
M. Tunç Arslan,
Onur Yorulmaz,
Erdinç L. Atılgan
Abstract:
Filters are the basic and most important blocks of most signal processing applications. In many applications, a group of parallel filters are used as filter banks. Parallel filter banks naturally require much more computations. Especially on chip applications, the resources are limited and shared among many algorithms. For this purpose, many filter optimization schemes are proposed to reduce the n…
▽ More
Filters are the basic and most important blocks of most signal processing applications. In many applications, a group of parallel filters are used as filter banks. Parallel filter banks naturally require much more computations. Especially on chip applications, the resources are limited and shared among many algorithms. For this purpose, many filter optimization schemes are proposed to reduce the number of resources that filtering operations require. In this work, a novel optimization algorithm is proposed to decrease the number of operations in a group of parallel filters. The filter coefficients are grouped in a two stage process which enables increased coefficient sharing between different filters. The algorithm is capable of decreasing the number of registers, look-up tables and DSP48s by up to 50\% of a regular parallel filter bank, without requiring increased sampling rate.
△ Less
Submitted 11 July, 2019;
originally announced July 2019.
-
Optimizing User Association and Activation Fractions in Heterogeneous Wireless Networks
Authors:
Vaibhav Singh,
Narayan Prasad,
Mustafa Y. Arslan,
Sampath Rangarajan
Abstract:
We consider the problem of maximizing the alpha-fairness utility over the downlink of a heterogeneous wireless network (HetNet) by jointly optimizing the association of users to transmission points (TPs) and the activation fractions of all TPs. Activation fraction of each TP is the fraction of the frame duration for which it is active, and together these fractions influence the interference seen i…
▽ More
We consider the problem of maximizing the alpha-fairness utility over the downlink of a heterogeneous wireless network (HetNet) by jointly optimizing the association of users to transmission points (TPs) and the activation fractions of all TPs. Activation fraction of each TP is the fraction of the frame duration for which it is active, and together these fractions influence the interference seen in the network. To address this joint optimization problem we adopt an approach wherein the activation fractions and the user associations are optimized in an alternating manner. The sub-problem of determining the optimal activation fractions is solved using an auxiliary function method that we show is provably convergent and is amenable to distributed implementation. On the other hand, the sub-problem of determining the user association is solved via a simple combinatorial algorithm. Meaningful performance guarantees are derived and a distributed variant offering identical guarantees is also proposed. The significant benefits of using the proposed algorithms are then demonstrated via realistic simulations.
△ Less
Submitted 22 March, 2015;
originally announced March 2015.
-
A Comparative Study of Scheduling Techniques for Multimedia Applications on SIMD Pipelines
Authors:
Mehmet Ali Arslan,
Flavius Gruian,
Krzysztof Kuchcinski
Abstract:
Parallel architectures are essential in order to take advantage of the parallelism inherent in streaming applications. One particular branch of these employ hardware SIMD pipelines. In this paper, we analyse several scheduling techniques, namely ad hoc overlapped execution, modulo scheduling and modulo scheduling with unrolling, all of which aim to efficiently utilize the special architecture desi…
▽ More
Parallel architectures are essential in order to take advantage of the parallelism inherent in streaming applications. One particular branch of these employ hardware SIMD pipelines. In this paper, we analyse several scheduling techniques, namely ad hoc overlapped execution, modulo scheduling and modulo scheduling with unrolling, all of which aim to efficiently utilize the special architecture design. Our investigation focuses on improving throughput while analysing other metrics that are important for streaming applications, such as register pressure, buffer sizes and code size. Through experiments conducted on several media benchmarks, we present and discuss trade-offs involved when selecting any one of these scheduling techniques.
△ Less
Submitted 26 February, 2015;
originally announced February 2015.
-
Information and Communication Technology Reputation for XU030 Quote Companies
Authors:
Sadi Evren Seker,
Bilal Cankir,
Mehmet Lutfi Arslan
Abstract:
By the increasing spread of information technology and Internet improvements, most of the large-scale companies are paying special attention to their reputation on many types of the information and communication technology. The increasing developments and penetration of new technologies into daily life, brings out paradigm shift on the perception of reputation and creates new concepts like esociet…
▽ More
By the increasing spread of information technology and Internet improvements, most of the large-scale companies are paying special attention to their reputation on many types of the information and communication technology. The increasing developments and penetration of new technologies into daily life, brings out paradigm shift on the perception of reputation and creates new concepts like esocieties, techno-culture and new media. Contemporary companies are trying to control their reputation over the new communities who are mostly interacting with social networks, web pages and electronic communication technologies. In this study, the reputation of top 30 Turkish companies, quoted to the Istanbul Stock Market, is studied, based on the information technology interfaces between company and society, such as social networks, blogs, wikis and web pages. The web reputation is gathered through 17 different parameters, collected from Google, Facebook, Twitter, Bing, Alexa, etc. The reputation index is calculated by z-index and fscoring formulations after the min-max normalization of each web reputation parameter.
△ Less
Submitted 16 June, 2014;
originally announced June 2014.
-
The Impact of Employment Web Sites' Traffic on Unemployment: A Cross Country Comparison
Authors:
M. Lutfi Arslan,
Sadi Evren Seker
Abstract:
Although employment web sites have recently become the main source for re- cruitment and selection process, the relation between those sites and unemploy- ment rates is seldom addressed. Deriving data from 32 countries and 427 web sites, this study explores the correlation between unemployment rates of European countries and the attractiveness of country specific employment web sites. It also comp…
▽ More
Although employment web sites have recently become the main source for re- cruitment and selection process, the relation between those sites and unemploy- ment rates is seldom addressed. Deriving data from 32 countries and 427 web sites, this study explores the correlation between unemployment rates of European countries and the attractiveness of country specific employment web sites. It also compares the changes in unemployment rates and traffic on all the aforementioned web sites. The results showed that there is a strong correlation between web sites traffic and unemployment rates.
△ Less
Submitted 8 March, 2014;
originally announced March 2014.
-
Approximate Computation of DFT without Performing Any Multiplications: Applications to Radar Signal Processing
Authors:
Alican Bozkurt,
Musa Tunç Arslan,
Rasim Akin Sevimli,
Cem Emre Akbas,
A. Enis Çetin
Abstract:
In many practical problems it is not necessary to compute the DFT in a perfect manner including some radar problems. In this article a new multiplication free algorithm for approximate computation of the DFT is introduced. All multiplications $(a\times b)$ in DFT are replaced by an operator which computes $sign(a\times b)(|a|+|b|)$. The new transform is especially useful when the signal processing…
▽ More
In many practical problems it is not necessary to compute the DFT in a perfect manner including some radar problems. In this article a new multiplication free algorithm for approximate computation of the DFT is introduced. All multiplications $(a\times b)$ in DFT are replaced by an operator which computes $sign(a\times b)(|a|+|b|)$. The new transform is especially useful when the signal processing algorithm requires correlations. Ambiguity function in radar signal processing requires high number of multiplications to compute the correlations. This new additive operator is used to decrease the number of multiplications. Simulation examples involving passive radars are presented.
△ Less
Submitted 3 February, 2014;
originally announced February 2014.
-
Web Based Reputation Index of Turkish Universities
Authors:
Mehmet Lutfi Arslan,
Sadi Evren Seker
Abstract:
This paper attempts to develop an online reputation index of Turkish universities through their online impact and effectiveness. Using 16 different web based parameters and employing normalization process of the results, we have ranked websites of Turkish universities in terms of their web presence. This index is first attempt to determine the tools of reputation of Turkish academic websites and w…
▽ More
This paper attempts to develop an online reputation index of Turkish universities through their online impact and effectiveness. Using 16 different web based parameters and employing normalization process of the results, we have ranked websites of Turkish universities in terms of their web presence. This index is first attempt to determine the tools of reputation of Turkish academic websites and would be a basis for further studies to examine the relation between reputation and the online effectiveness of the universities.
△ Less
Submitted 25 January, 2014;
originally announced January 2014.
