• Korean Journal of Artificial Intelligence
• /
• v.12 no.2
• /
• pp.9-16
• /
• 2024

In this study, we delve into the effects of personalization algorithms on the creation of "filter bubbles," which can isolate individuals intellectually by reinforcing their pre-existing biases, particularly through personalized Google searches. By setting up accounts with distinct ideological learnings-progressive and conservative-and employing deep neural networks to simulate user interactions, we quantitatively confirmed the existence of filter bubbles. Our investigation extends to the deployment of an LSTM model designed to assess political orientation in text, enabling us to bias accounts deliberately and monitor their increasing ideological inclinations. We observed politically biased search results appearing over time in searches through biased accounts. Additionally, the political bias of the accounts continued to increase. These results provide numerical evidence for the existence of filter bubbles and demonstrate that these bubbles exert a greater influence on search results over time. Moreover, we explored potential solutions to mitigate the influence of filter bubbles, proposing methods to promote a more diverse and inclusive information ecosystem. Our findings underscore the significance of filter bubbles in shaping users' access to information and highlight the urgency of addressing this issue to prevent further political polarization and media habit entrenchment. Through this research, we contribute to a broader understanding of the challenges posed by personalized digital environments and offer insights into strategies that can help alleviate the risks of intellectual isolation caused by filter bubbles.

• Journal of Advanced Navigation Technology
• /
• v.6 no.2
• /
• pp.138-150
• /
• 2002

This paper addresses the problem of combining adaptive polarization processing and space-time processing for further performance improvement of radar target detection in clutter and Jammer environments. Since the most straightforward cascade combinations have quite limited performance improvement potentials, we focus on the development of adaptive processing in the joint polarization-space-time domain. Unlike a direct extension of some existing space-time processing algorithms to the joint domain, the processing algorithm developed in this paper does not need a potentially costly polarization filter bank to cover the unknown target polarization parameter. The performance of the new algorithm is derived and evaluated in terms of the probability of detection and the probability of false alarm, and it is compared with other algorithms that do not utilize the polarization information or assume that the target polarization is known.

• Geophysics and Geophysical Exploration
• /
• v.12 no.3
• /
• pp.246-254
• /
• 2009

We developed a novel polarization filter to separate linearly polarized waves from elliptically polarized waves in an infinite homogeneous medium and at the free surface using methods of multicomponent complex trace analysis. Sensitivity to filter parameters were examined using synthetic data simulating particle motions in a homogeneous medium. For known amplitude ratios of horizontal-to-vertical components of P and Rayleigh waves $C_L$ and $C_R$, respectively, the polarization filter precisely removes Rayleigh waves. Errors in the vertical and horizontal components of the filtered results increase with the ratio $C_R$/$C_L$ and the product $C_R$.$C_L$, respectively. The vertical component errors also increase rapidly as the ratios of applied-to-modeled values of $C_L$ and $C_R$ ($C_L'$/$C_L$ and $C_R'$/$C_R$) decrease, and are sensitive to $C_R'$/$C_R$ and $C_L'$/$C_L$ for small and large incidence angles, respectively. Errors of the filter are exactly the same for shear waves when the incidence angle is the supplementary of P-wave incidence angle.

• Journal of electromagnetic engineering and science
• /
• v.15 no.4
• /
• pp.206-212
• /
• 2015

This paper introduces new wideband antenna-filter-antenna (AFA) uniform arrays that can be utilized as frequency selective surfaces (FSS) with low loss and sharp roll-off response, which are highly desirable characteristics for millimeter wave applications. The design adopts a simple 3-layer single polarization structure consisting of two patch antennas and a resonator. Both simulations and measurements are used to characterize the performances of the proposed design. Overall results show 18.5% 10-dB bandwidth. For the targeted band the insertion loss is less than 0.2 dB. Possible applications include quasi-optical amps, grid mixers and radomes for aircraft radar antennas.

• Journal of the Optical Society of Korea
• /
• v.20 no.1
• /
• pp.180-187
• /
• 2016

The independent operation of a color filter for incident polarization is demonstrated using a guided-mode resonance (GMR) filter employing a metal-insulator-metal (MIM) waveguide. To achieve independent operation, a rectangular MIM grating is proposed as a wave-guide resonator. The design considerations are discussed and include how to determine the grating period and slit width. Power flow distribution is observed with slit width variation. Blue-green, green-red, and blue-red filters for corresponding x- and y-polarizations are demonstrated as application examples with numerical simulation with rectangle-shaped MIM grating. As a practical application, feasibility as a chromatic polarizer is discussed.

• Proceedings of the Optical Society of Korea Conference
• /
• 2006.07a
• /
• pp.55-56
• /
• 2006

• Journal of Korea Multimedia Society
• /
• v.21 no.3
• /
• pp.369-381
• /
• 2018

Nowadays a personalization algorithm is gaining huge attention. It gives users selective information which is helpful and interesting in a deluge of information based on their past behavior on the internet. However there is also a fatal side effect that the user can only get restricted information on restricted topics selected by the algorithm. Basically, the personalization algorithm makes users have a narrower perspective and even stronger bias because users have less chances to get views of opponent. Eli Pariser called this problem the 'filter bubble' in his book. It is important to understand exactly what a filter bubble is to solve the problem. Therefore, this paper shows how much Google's personalized search algorithm influences search result through an experiment with deep neural networks acting like users. At the beginning of the experiment, two Google accounts are newly created, not to be influenced by the Google's personalized search algorithm. Then the two pure accounts get politically biased by two methods. We periodically calculate the numerical score depending on the character of links and it shows how biased the account is. In conclusion, this paper shows the formation process of filter bubble by a personalization algorithm through the experiment.

• Journal of IKEEE
• /
• v.26 no.3
• /
• pp.416-421
• /
• 2022

In this study, we present a fully automated system that combines camera technology with liquid crystal technology to create a polarization camera capable of detecting the partial linear polarization of light reflected from an object. The use of twisted nematic (TN) liquid crystals that electro-optically modulate the polarization plane of light eliminates the need to mechanically rotate the polarizing filter in front of the camera lens. Images obtained using these techniques are imaged by computer software. In addition, liquid crystal panels have been produced in a square shape, but many camera lenses are usually round, and lighting or other driving units are installed around the lens, so space is optimized through the application of a circular liquid crystal display. Through the development of this technology, an electrically switchable and space-optimized liquid crystal polarizer is developed.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2017.06a
• /
• pp.102-104
• /
• 2017

최근 Convolutional neural network (CNN)을 도입하여, SAR 영상의 목표물 인식 알고리즘이 높은 성능을 보여주었다. SAR 영상은 4 종류의 polarization 정보로 구성되어있다. 기계와 신호처리의 비용으로 인하여 일부 데이터는 적은 수의 polarization 정보를 가지고 있다. 따라서 우리는 SAR 영상 data 를 멀티모달 데이터로 해석하였다. 그리고 우리는 이러한 멀티모달 데이터에 잘 작동할 수 있는 콘볼루션 신경망을 제안하였다. 우리는 데이터가 포함하는 모달의 수에 반비례 하도록 scale factor 구성하고 이를 입력 크기조절에 사용하였다. 입력의 크기를 조절하여, 네트워크는 특징맵의 크기를 모달의 수와 상관없이 일정하게 유지할 수 있었다. 또한 제안하는 입력 크기조절 방법은 네트워크의 dead filter 의 수를 감소 시켰고, 이는 네트워크가 자신의 capacity 를 잘 활용한다는 것을 의미한다. 또 제안된 네트워크는 특징맵을 구성할 때 다양한 모달을 활용하였고, 이는 네트워크가 모달간의 상관관계를 학습했다는 것을 의미한다. 그 결과, 제안된 네트워크의 성능은 입력 크기조절이 없는 일반적인 네트워크보다 높은 성능을 보여주었다. 또한 우리는 전이학습의 개념을 이용하여 네트워크를 모달의 수가 많은 데이터부터 차례대로 학습시켰다. 전이학습을 통하여 네트워크가 학습되었을 때, 제안된 네트워크는 특정 모달의 조합 경우만을 위해 학습된 네트워크보다 높은 성능을 보여준다.

• Korean Journal of Optics and Photonics
• /
• v.9 no.4
• /
• pp.270-273
• /
• 1998

We have demonstrated the wavelength tunable ultrashort pulse fiber ring laser proposed the novel method for measuring the cavity loss from relaxation oscillation frequency and the pump power. To suppress the polarization instability the laser cavity is configured with polarization maintaining fibers and to control the center wavelength a 2.4 nm bandwidth tunable wavelength filter was inserted in the cavity. The laser has 8 picosecond pulse width, 10 GHz repetition rate and 1.2 mW average power in 1530-1560 nm operation range.