Pub Date : 2022-05-24DOI: 10.1109/NCC55593.2022.9806756
P. Barik, R. Datta
For efficient spectrum utilization, the licensed band of cellular network is used for Device-to-Device (D2D) commu-nication. Reuse of resource blocks (RBs) in D2D links creates interference to cellular users (CUs) and the interference becomes severe when D2D transmitters and receivers are connected with different cells. However, minimization of both intracell and inter-cell interference is a challenging task while realizing intercell D2D communication. In this paper, we propose a resource allocation scheme, named CoRAL, where the neighbor base stations (BSs) coordinate between themselves to minimize the interference caused by the D2D communication. Each base station computes a look-up table (L UT), comprising the degradation in throughput for each RB. The coordination between the serving BS (S-eNB) and its neighbor BSs (N-eNBs) is obtained by sharing the LUTs. The S-eNB combines all the LUTs and forms an updated table, which is used for the resource allocation algorithm. We formulate the resource allocation problem as a minimization problem of the degradation of achieved throughput due to sharing of RBs between D2D users and CUs in the system. CoRAL enhances the system throughput by minimizing the intra and intercell interference simultaneously in the presence of D2D communication. It uses an iterative approach for allocating RBs to D2D users. The simulation results show that CoRAL outperforms the uncoordinated resource allocation techniques where a single base station decides the resource assignment procedure and a popular scheme GALLERY from existing literature.
{"title":"CoRAL: Coordinated Resource Allocation for Intercell D2D Communication in Cellular Networks","authors":"P. Barik, R. Datta","doi":"10.1109/NCC55593.2022.9806756","DOIUrl":"https://doi.org/10.1109/NCC55593.2022.9806756","url":null,"abstract":"For efficient spectrum utilization, the licensed band of cellular network is used for Device-to-Device (D2D) commu-nication. Reuse of resource blocks (RBs) in D2D links creates interference to cellular users (CUs) and the interference becomes severe when D2D transmitters and receivers are connected with different cells. However, minimization of both intracell and inter-cell interference is a challenging task while realizing intercell D2D communication. In this paper, we propose a resource allocation scheme, named CoRAL, where the neighbor base stations (BSs) coordinate between themselves to minimize the interference caused by the D2D communication. Each base station computes a look-up table (L UT), comprising the degradation in throughput for each RB. The coordination between the serving BS (S-eNB) and its neighbor BSs (N-eNBs) is obtained by sharing the LUTs. The S-eNB combines all the LUTs and forms an updated table, which is used for the resource allocation algorithm. We formulate the resource allocation problem as a minimization problem of the degradation of achieved throughput due to sharing of RBs between D2D users and CUs in the system. CoRAL enhances the system throughput by minimizing the intra and intercell interference simultaneously in the presence of D2D communication. It uses an iterative approach for allocating RBs to D2D users. The simulation results show that CoRAL outperforms the uncoordinated resource allocation techniques where a single base station decides the resource assignment procedure and a popular scheme GALLERY from existing literature.","PeriodicalId":403870,"journal":{"name":"2022 National Conference on Communications (NCC)","volume":"56 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114043156","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-24DOI: 10.1109/NCC55593.2022.9806728
Moyukh Laha, R. Datta
In the next generation of vehicular network applications, complex data processing and reliable and quick message transmissions are critical. Traditional cellular macro base stations and IEEE WAVE technology are incapable of supporting such high data speeds and ultra-reliable low latency communication. The combination of 5G RSUs equipped with mmWave beams (mmRSUs) and edge computing methods have been proposed as a possible solution for meeting such service needs. However, since urban vehicle traffic is often predictable, the mmRSUs need not be kept ON all the time to provide services. Instead, the mmRSUs may be dynamically turned ON/OFF depending on current traffic conditions, hence reducing energy consumption without compromising service. We construct the intelligent switching of mmRSUs as an Integer Linear Program to maximize the system's utility by dynamically turning them on/off in order to spend less energy. We propose a strategy based on Deep Q-Learning to accomplish the goal and demonstrate its usefulness in a city with real traffic.
{"title":"Intelligent On/Off Switching of mmRSUs in Urban Vehicular Networks: A Deep Q-Learning Approach","authors":"Moyukh Laha, R. Datta","doi":"10.1109/NCC55593.2022.9806728","DOIUrl":"https://doi.org/10.1109/NCC55593.2022.9806728","url":null,"abstract":"In the next generation of vehicular network applications, complex data processing and reliable and quick message transmissions are critical. Traditional cellular macro base stations and IEEE WAVE technology are incapable of supporting such high data speeds and ultra-reliable low latency communication. The combination of 5G RSUs equipped with mmWave beams (mmRSUs) and edge computing methods have been proposed as a possible solution for meeting such service needs. However, since urban vehicle traffic is often predictable, the mmRSUs need not be kept ON all the time to provide services. Instead, the mmRSUs may be dynamically turned ON/OFF depending on current traffic conditions, hence reducing energy consumption without compromising service. We construct the intelligent switching of mmRSUs as an Integer Linear Program to maximize the system's utility by dynamically turning them on/off in order to spend less energy. We propose a strategy based on Deep Q-Learning to accomplish the goal and demonstrate its usefulness in a city with real traffic.","PeriodicalId":403870,"journal":{"name":"2022 National Conference on Communications (NCC)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114236019","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-24DOI: 10.1109/NCC55593.2022.9806721
G. Mishra, H. Rath, S. Nadaf
Real-time data generation and collection to analyse the network performance is difficult for large-scale networks having limited accessibility. In this paper we propose a framework which can provide realistic si/e-mulations, and generate synthetic data closer to real-time data that replaces the traditionally used deterministic and probabilistic models. This framework uses an emulation based platform to replicate real network scenarios. The emulator acts as a base layer with necessary APIs to enable customized inclusion of analytics services in a plug-and-play manner through the framework. This framework can be used to acquire data required for different Machine Learning (ML) models in order to reduce costly and time-consuming data collection effort in network analytics.
{"title":"Emulation as a Service (EaaS): A Plug-n-Play Framework for Benchmarking Network Analytics","authors":"G. Mishra, H. Rath, S. Nadaf","doi":"10.1109/NCC55593.2022.9806721","DOIUrl":"https://doi.org/10.1109/NCC55593.2022.9806721","url":null,"abstract":"Real-time data generation and collection to analyse the network performance is difficult for large-scale networks having limited accessibility. In this paper we propose a framework which can provide realistic si/e-mulations, and generate synthetic data closer to real-time data that replaces the traditionally used deterministic and probabilistic models. This framework uses an emulation based platform to replicate real network scenarios. The emulator acts as a base layer with necessary APIs to enable customized inclusion of analytics services in a plug-and-play manner through the framework. This framework can be used to acquire data required for different Machine Learning (ML) models in order to reduce costly and time-consuming data collection effort in network analytics.","PeriodicalId":403870,"journal":{"name":"2022 National Conference on Communications (NCC)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115052502","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-24DOI: 10.1109/NCC55593.2022.9806748
A. Kherani, T. V. Sreejith
We consider a large floor that has RIS elements assisting the transmissions from the different terminal devices towards a common access point located at the center. Showing that the near-field model is the appropriate model for this system, we provide an approach to study the improvement in performance when using RIS in a system that utilizes Successive Interference Cancellation (SIC), along with Random Access, like Slotted ALOHA. Without the use of RIS, the possible spatial region where SIC could have been used is very small, and introduction of RIS provides a promising improvement. Partial RIS sharing with SIC is seen to offer a significant further improvement in this region. By finding the optimal channel access probabilities by viewing the end-devices as individual players in a game with a common utility function, we see that the overall average system throughput performance improves with RIS assistance.
{"title":"On RIS-Assisted Random Access Systems with Successive Interference Cancellation","authors":"A. Kherani, T. V. Sreejith","doi":"10.1109/NCC55593.2022.9806748","DOIUrl":"https://doi.org/10.1109/NCC55593.2022.9806748","url":null,"abstract":"We consider a large floor that has RIS elements assisting the transmissions from the different terminal devices towards a common access point located at the center. Showing that the near-field model is the appropriate model for this system, we provide an approach to study the improvement in performance when using RIS in a system that utilizes Successive Interference Cancellation (SIC), along with Random Access, like Slotted ALOHA. Without the use of RIS, the possible spatial region where SIC could have been used is very small, and introduction of RIS provides a promising improvement. Partial RIS sharing with SIC is seen to offer a significant further improvement in this region. By finding the optimal channel access probabilities by viewing the end-devices as individual players in a game with a common utility function, we see that the overall average system throughput performance improves with RIS assistance.","PeriodicalId":403870,"journal":{"name":"2022 National Conference on Communications (NCC)","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133860347","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-24DOI: 10.1109/NCC55593.2022.9806799
Giridhar Pamisetty, S. Kodukula
Automatic dubbing or machine dubbing is the process of replacing the speech in the source video with the desired language speech, which is synthesized using a text-to-speech synthesis (TTS) system. The synthesized speech should align with the events in the source video to have a realistic experience. Most of the existing prosodic alignment processes operate on the synthesized speech by controlling the speaking rate. In this paper, we propose subtitle synthesis, a unified approach for the prosodic alignment that operates at the feature level. Modifying the prosodic parameters at the feature level will not degrade the naturalness of the synthesized speech. We use both inter and intra utterance alignment in the prosodic alignment process. We should have control over the duration of the phonemes to perform alignment at the feature level to achieve synchronization between the synthesized and the source speech. So, we use the Prosody-TTS system to synthesize the speech, which has the provision to control the duration of phonemes and fundamental frequency (f0) during the synthesis. The subjective evaluation of the translated audiovisual content (lecture videos) resulted in a mean opinion score (MOS) of 4.104 that indicates the effectiveness of the proposed prosodic alignment process.
{"title":"Subtitle Synthesis using Inter and Intra utterance Prosodic Alignment for Automatic Dubbing","authors":"Giridhar Pamisetty, S. Kodukula","doi":"10.1109/NCC55593.2022.9806799","DOIUrl":"https://doi.org/10.1109/NCC55593.2022.9806799","url":null,"abstract":"Automatic dubbing or machine dubbing is the process of replacing the speech in the source video with the desired language speech, which is synthesized using a text-to-speech synthesis (TTS) system. The synthesized speech should align with the events in the source video to have a realistic experience. Most of the existing prosodic alignment processes operate on the synthesized speech by controlling the speaking rate. In this paper, we propose subtitle synthesis, a unified approach for the prosodic alignment that operates at the feature level. Modifying the prosodic parameters at the feature level will not degrade the naturalness of the synthesized speech. We use both inter and intra utterance alignment in the prosodic alignment process. We should have control over the duration of the phonemes to perform alignment at the feature level to achieve synchronization between the synthesized and the source speech. So, we use the Prosody-TTS system to synthesize the speech, which has the provision to control the duration of phonemes and fundamental frequency (f0) during the synthesis. The subjective evaluation of the translated audiovisual content (lecture videos) resulted in a mean opinion score (MOS) of 4.104 that indicates the effectiveness of the proposed prosodic alignment process.","PeriodicalId":403870,"journal":{"name":"2022 National Conference on Communications (NCC)","volume":"66 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122290152","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-24DOI: 10.1109/NCC55593.2022.9806769
Ayush Agarwal, Amitabh Swain, Jagabandhu Mishra, S. Prasanna
Privacy is the major concern that comes to the user's mind before sharing their data. There are various methods proposed in literature for providing privacy to speech data. Previous works that have been done to protect the speaker identity were done for speech applications like automatic speech recognition (ASR), speech analysis, etc. For these applications the presence of speaker identity is not essential while processing. The objective of this work is to provide privacy to the task in which presence of speaker identity is essential at the time of processing. In this work, privacy is provided to the speaker identity information present in speech signals while performing automatic speaker verification (ASV) tasks. In order to achieve the same, this work proposes a prosody modification based approach. The proposed approach is able to conceal the speaker identity from human perception by changing the pitch of the speech utterances with a pitch modification factor of $alphageq 1$ But at the same time the ASV system provides consistent performance irrespective of the change in pitch (i.e. for $alphageq 1)$. The same evidence has been shown through experiments in TIMIT and IITG-MV databases. A subjective study has also performed to verify the extent of speaker anonymization with respect to humans. The subjective study evaluates the performance in terms of mean opinion score (MOS). The observed MOS signifies the ability of the proposed approach to conceal the speaker's identity.
{"title":"Significance of Prosody Modification in Privacy Preservation on speaker verification","authors":"Ayush Agarwal, Amitabh Swain, Jagabandhu Mishra, S. Prasanna","doi":"10.1109/NCC55593.2022.9806769","DOIUrl":"https://doi.org/10.1109/NCC55593.2022.9806769","url":null,"abstract":"Privacy is the major concern that comes to the user's mind before sharing their data. There are various methods proposed in literature for providing privacy to speech data. Previous works that have been done to protect the speaker identity were done for speech applications like automatic speech recognition (ASR), speech analysis, etc. For these applications the presence of speaker identity is not essential while processing. The objective of this work is to provide privacy to the task in which presence of speaker identity is essential at the time of processing. In this work, privacy is provided to the speaker identity information present in speech signals while performing automatic speaker verification (ASV) tasks. In order to achieve the same, this work proposes a prosody modification based approach. The proposed approach is able to conceal the speaker identity from human perception by changing the pitch of the speech utterances with a pitch modification factor of $alphageq 1$ But at the same time the ASV system provides consistent performance irrespective of the change in pitch (i.e. for $alphageq 1)$. The same evidence has been shown through experiments in TIMIT and IITG-MV databases. A subjective study has also performed to verify the extent of speaker anonymization with respect to humans. The subjective study evaluates the performance in terms of mean opinion score (MOS). The observed MOS signifies the ability of the proposed approach to conceal the speaker's identity.","PeriodicalId":403870,"journal":{"name":"2022 National Conference on Communications (NCC)","volume":"51 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128772993","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-24DOI: 10.1109/NCC55593.2022.9806765
V. C. Aparna, A. Gandhi, S. BhaskaraNaik, R. Harsh
The main objective of this paper is to study two types of minimally invasive coaxial antennas for microwave ablation served for the treatment of liver cancer. Existing microwave antenna designs for ablation are around 2 mm and above in outer diameter which requires a greater hole size in the human body for the insertion of the applicator into the site of tumor. In this paper, we propose miniaturized coaxial slot antennas with an outer diameter of 1 mm which satisfies the medical requirement to minimize the invasiveness. Two types of miniaturized antennas namely monopole with slot and dual slot with an outer diameter of 1 mm are designed and simulated in CST at two different frequencies (915 MHz and 2.45 GHz) to determine the variation in the characteristics of both antennas w.r.t. frequency. Return Loss, Voltage Standing Wave Ratio (VSWR) and Specific Absorption Rate (SAR) pattern of antennas are analyzed to determine the efficiency of the antenna, heating pattern and ablation zone. Simulation results indicate that a dual slot antenna has uniform power distribution around the region of tumor, minimum backward heating and better sphericity in ablation pattern compared to a monopole slot antenna. Also, the antennas operated at 2.45 G Hz are more suitable for creating rapid spherical ablation with a larger diameter compared to the antennas operated at 915 MHz. So we can conclude that a dual slot antenna operated at 2.45 GHz is best suited for microwave ablation.
本文的主要目的是研究两种用于微波消融治疗肝癌的微创同轴天线。现有的消融术微波天线的外径约为2毫米及以上，这需要在人体上有更大的孔尺寸，以便将应用器插入肿瘤部位。在本文中，我们提出了外径为1毫米的小型化同轴缝隙天线，以满足医疗要求，以减少侵入性。设计了外径为1 mm的单极槽和双槽两种小型化天线，并在CST中对915 MHz和2.45 GHz两种不同频率下的天线进行了仿真，以确定两种天线的wrt频率特性变化。分析天线的回波损耗、电压驻波比(VSWR)和比吸收率(SAR)方向图，确定天线的效率、加热方向图和烧蚀区。仿真结果表明，与单极槽天线相比，双槽天线在肿瘤周围具有均匀的功率分布、最小的反向加热和更好的烧蚀方向球度。此外，与工作在915 MHz的天线相比，工作在2.45 G Hz的天线更适合产生直径更大的快速球形烧蚀。因此，工作频率为2.45 GHz的双缝隙天线最适合微波消融。
{"title":"Minimally Invasive Microwave Ablation Antenna Designs at 915 MHz and 2.45 GHz","authors":"V. C. Aparna, A. Gandhi, S. BhaskaraNaik, R. Harsh","doi":"10.1109/NCC55593.2022.9806765","DOIUrl":"https://doi.org/10.1109/NCC55593.2022.9806765","url":null,"abstract":"The main objective of this paper is to study two types of minimally invasive coaxial antennas for microwave ablation served for the treatment of liver cancer. Existing microwave antenna designs for ablation are around 2 mm and above in outer diameter which requires a greater hole size in the human body for the insertion of the applicator into the site of tumor. In this paper, we propose miniaturized coaxial slot antennas with an outer diameter of 1 mm which satisfies the medical requirement to minimize the invasiveness. Two types of miniaturized antennas namely monopole with slot and dual slot with an outer diameter of 1 mm are designed and simulated in CST at two different frequencies (915 MHz and 2.45 GHz) to determine the variation in the characteristics of both antennas w.r.t. frequency. Return Loss, Voltage Standing Wave Ratio (VSWR) and Specific Absorption Rate (SAR) pattern of antennas are analyzed to determine the efficiency of the antenna, heating pattern and ablation zone. Simulation results indicate that a dual slot antenna has uniform power distribution around the region of tumor, minimum backward heating and better sphericity in ablation pattern compared to a monopole slot antenna. Also, the antennas operated at 2.45 G Hz are more suitable for creating rapid spherical ablation with a larger diameter compared to the antennas operated at 915 MHz. So we can conclude that a dual slot antenna operated at 2.45 GHz is best suited for microwave ablation.","PeriodicalId":403870,"journal":{"name":"2022 National Conference on Communications (NCC)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129886212","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-24DOI: 10.1109/NCC55593.2022.9806739
Komal Ojha, K. Appaiah
Multiplexing using principal modes (PMs) of mul-timode fibers is known to be effective at enhancing data rates for channel variations that are up to first order in modulation bandwidth $Omega$. However, for large $Omega$, the second order group delay terms (dependent on $Omega^{2}$) diminish the effectiveness of PMs. In this paper, we show that simultaneously feeding back PMs as well as second order group delay matrices permits transmission that overcomes the higher order modal dispersion. In addition, using manifold based quantization of PMs minimizes feedback overheads. Simulations reveal that modal dispersion is effectively eliminated with very little additional feedback overhead.
{"title":"Feedback based Compensation of Second Order Modal Dispersion in Principal Mode based Multiplexed MMF Links","authors":"Komal Ojha, K. Appaiah","doi":"10.1109/NCC55593.2022.9806739","DOIUrl":"https://doi.org/10.1109/NCC55593.2022.9806739","url":null,"abstract":"Multiplexing using principal modes (PMs) of mul-timode fibers is known to be effective at enhancing data rates for channel variations that are up to first order in modulation bandwidth $Omega$. However, for large $Omega$, the second order group delay terms (dependent on $Omega^{2}$) diminish the effectiveness of PMs. In this paper, we show that simultaneously feeding back PMs as well as second order group delay matrices permits transmission that overcomes the higher order modal dispersion. In addition, using manifold based quantization of PMs minimizes feedback overheads. Simulations reveal that modal dispersion is effectively eliminated with very little additional feedback overhead.","PeriodicalId":403870,"journal":{"name":"2022 National Conference on Communications (NCC)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117072145","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-24DOI: 10.1109/NCC55593.2022.9806747
Shajahan Aboobacker, Akash Verma, Deepu Vijayasenan, Sumam David S., P. Suresh, Saraswathy Sreeram
Automation in the detection of malignancy in effusion cytology helps to save time and workload for cytopathologists. Cytopathologists typically consider a low-resolution image to identify the malignant regions. The identified regions are scanned at a higher resolution to confirm malignancy by investigating the cell level behaviour. Scanning and processing time can be saved by zooming only the identified malignant regions instead of entire low-resolution images. This work predicts malignancy in cytology images at a very low resolution (4X). Annotation of cytology images at a very low resolution is challenging due to the blurring of features such as nuclei and texture. We address this issue by upsampling the very low-resolution images using adversarial training. This work develops a semantic segmentation model trained on 10X images and reuse the network to utilize the 4X images. The prediction results of low resolution images improved by 15% in average F-score for adversarial based upsampling compared to a bicubic filter. The high resolution model gives a 95% average F-score for high resolution images. Also, the sub-area of the whole slide that requires to be scanned at high magnification is reduced by approximately 61% while using adversarial based upsampling compared to a bicubic filter.
{"title":"Semantic Segmentation on Low Resolution Cytology Images of Pleural and Peritoneal Effusion","authors":"Shajahan Aboobacker, Akash Verma, Deepu Vijayasenan, Sumam David S., P. Suresh, Saraswathy Sreeram","doi":"10.1109/NCC55593.2022.9806747","DOIUrl":"https://doi.org/10.1109/NCC55593.2022.9806747","url":null,"abstract":"Automation in the detection of malignancy in effusion cytology helps to save time and workload for cytopathologists. Cytopathologists typically consider a low-resolution image to identify the malignant regions. The identified regions are scanned at a higher resolution to confirm malignancy by investigating the cell level behaviour. Scanning and processing time can be saved by zooming only the identified malignant regions instead of entire low-resolution images. This work predicts malignancy in cytology images at a very low resolution (4X). Annotation of cytology images at a very low resolution is challenging due to the blurring of features such as nuclei and texture. We address this issue by upsampling the very low-resolution images using adversarial training. This work develops a semantic segmentation model trained on 10X images and reuse the network to utilize the 4X images. The prediction results of low resolution images improved by 15% in average F-score for adversarial based upsampling compared to a bicubic filter. The high resolution model gives a 95% average F-score for high resolution images. Also, the sub-area of the whole slide that requires to be scanned at high magnification is reduced by approximately 61% while using adversarial based upsampling compared to a bicubic filter.","PeriodicalId":403870,"journal":{"name":"2022 National Conference on Communications (NCC)","volume":"272 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122831325","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-05-24DOI: 10.1109/NCC55593.2022.9806811
Vidit Kumar
The majority of videos in the internet lack semantic tags, which complicates indexing and retrieval, and mandates the adoption of critical content-based analysis approaches. Earlier works relies on hand-crafted features, which hardly represents the temporal dynamics. Later, video representations learned through supervised deep learning methods were found to be effective, but at the cost of large labeled dataset. Recently, self-supervised based methods for video representation learning are proposed within the community to harness the freely available unlabeled videos. However, most of these methods are based on single pretext task, which limits the learning of generalizable representations. This work proposes to leverage multiple pretext tasks to enhance video representation learning and generalizability. We jointly optimized the C3D network by using multiple pretext tasks such as: rotation prediction, speed prediction, time direction prediction and instance discrimination. The nearest neighbour task is used to analyze the learned features. And for action recognition task, the network is further fine-tuned with pretrained weights. We use the UCF-101 dataset for the experiments and, achieves 28.45% retrieval accuracy (Recall@l), and 68.85% fine-tuned action recognition accuracy, which is better than state-of-the-arts.
{"title":"Unsupervised Learning of Spatio-Temporal Representation with Multi-Task Learning for Video Retrieval","authors":"Vidit Kumar","doi":"10.1109/NCC55593.2022.9806811","DOIUrl":"https://doi.org/10.1109/NCC55593.2022.9806811","url":null,"abstract":"The majority of videos in the internet lack semantic tags, which complicates indexing and retrieval, and mandates the adoption of critical content-based analysis approaches. Earlier works relies on hand-crafted features, which hardly represents the temporal dynamics. Later, video representations learned through supervised deep learning methods were found to be effective, but at the cost of large labeled dataset. Recently, self-supervised based methods for video representation learning are proposed within the community to harness the freely available unlabeled videos. However, most of these methods are based on single pretext task, which limits the learning of generalizable representations. This work proposes to leverage multiple pretext tasks to enhance video representation learning and generalizability. We jointly optimized the C3D network by using multiple pretext tasks such as: rotation prediction, speed prediction, time direction prediction and instance discrimination. The nearest neighbour task is used to analyze the learned features. And for action recognition task, the network is further fine-tuned with pretrained weights. We use the UCF-101 dataset for the experiments and, achieves 28.45% retrieval accuracy (Recall@l), and 68.85% fine-tuned action recognition accuracy, which is better than state-of-the-arts.","PeriodicalId":403870,"journal":{"name":"2022 National Conference on Communications (NCC)","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130638377","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：