Abstract

Performing sound source separation and visual object seg- mentation jointly in naturally occurring videos is a notori- ously difficult task, especially in the absence of annotated data. In this study, we leverage the concurrency between audio and visual modalities in an attempt to solve the joint audio-visual segmentation problem in a self-supervised man- ner. Human beings interact with the physical world through a few sensory systems such as vision, auditory, movement, etc. The usefulness of the interplay of such systems lies in the con- cept of degeneracy [1]. It tells us that the cross-modal signals can educate each other without the presence of an external supervisor. In this work, we efficiently exploit this fact that learning from one modality inherently helps to find patterns in others by introducing a novel audio-visual fusion technique. Also, to the best of our knowledge, we are the first to ad- dress the partially occluded sound source segmentation task. Our study shows that the proposed model significantly out- performs existing state-of-the-art methods in both visual and audio source separation tasks

Abstract

Estimation of time delay from the received broadband signals like speech, collected at two or more spatially distributed microphones, has many applications. Methods like the cross-correlation of the signals directly and generalized crosscorrelation based methods (GCC and GCC-PHAT) have been used for several years to estimate the time delay. Performance of these methods degrades due to noise, multi-path reflections, and reverberation in a practical environment, like a live room. The estimated time delay is usually robust due to the averaging effect of the delay obtained over several frames in an utterance of a few seconds. The robustness is affected if the varying time delay of a moving speaker is desired. A smaller duration for averaging results in errors in the estimation of the time delay, and a longer duration for averaging results in loss of accuracy. Since the single frequency filtering (SFF) based analysis provides an estimation of the instantaneous time delay, it is possible to study the trade off between accuracy and robustness. This paper examines this trade-off in determining the number of stationary speakers from mixed signals and in tracking a speaker moving along a straight line path and along a circular path. The results are illustrated for actual data collected in a live room.

Abstract

Humans overcome the degradation effect of background noise on the speech by changing their vocal characteristics dynamically, which text-to-speech systems trained on clean speech cannot, re- sulting in degraded intelligibility. To improve the intelligibility of such system requires a large amount of speech samples, which is difficult to collect for various conditions like noise backgrounds and signal-to-noise ratios. This paper presents a noise dependent enhancement to text-to-speech, based on generative adversarial network training framework to generate intelligible speech in noise. The learning mechanism for the synthesizer network is inspired from the acoustic feedback humans use to nullify the effect of vari- ous background noises. The system thus trained is evaluated under cafeteria noise condition with two objective measures, which indi- cated improvement in intelligibility compared to models trained on clean speech across 3 SNRs. The proposed modification does not require any additional training data and can be applied to a variety of Deep Neura

Abstract

We propose an all-CMOS, supply independent, second-order temperature compensated, process invariant, 10.09 μA current reference circuit working on a 1.8 V power supply. The current reference is built using an adaptive regulated cascode configuration integrated with a Widlar current source. It is implemented in 180 nm UMC MPW RF process and consumes an area of 0.014 mm 2 . It has a very low line sensitivity of 0.52 %/V and a temperature coefficient of 9.94 ppm/°C over the automotive temperature range of −40°C to 125°C. The proposed current reference is used to build a high quality, power efficient stable ring oscillator circuit working at 377.9 MHz frequency to be used in IoT SoCs. The ring oscillator shows a variation of less than 1.1 % in frequency, with supply voltage over the range 1.5 V to 2 V. The invariance of frequency with supply voltage can be exploited to reduce the power consumption up to 36.93 % in the oscillator circuit.

Abstract

The exponential rise of online social media has enabled the creation, distribution, and consumption of information at an unprecedented rate. However, it has also led to the burgeoning of various forms of online abuse. Increasing cases of online antisemitism have be- come one of the major concerns because of its socio-political con- sequences. Unlike other major forms of online abuse like racism, sexism, etc., online antisemitism has not been studied much from a machine learning perspective. To the best of our knowledge, we present the first work in the direction of automated multimodal de- tection of online antisemitism. The task poses multiple challenges that include extracting signals across multiple modalities, contex- tual references, and handling multiple aspects of antisemitism. Un- fortunately, there does not exist any publicly available benchmark corpus for this critical task. Hence, we collect and label two datasets with 3,102 and 3,509 social media posts from Twitter and Gab re- spectively. Further, we present a multimodal deep learning system that detects the presence of antisemitic content and its specific anti- semitism category using text and images from posts. We perform an extensive set of experiments on the two datasets to evaluate the ef- ficacy of the proposed system. Finally, we also present a qualitative analysis of our study.

Abstract

Social media has grown exponentially in a short period, coming to the forefront of communications and online interactions. Despite their rapid growth, social media platforms have been unable to scale to different languages globally and remain inaccessible to many. In this paper, we characterize Koo, a multilingual micro-blogging site that rose in popularity in 2021, as an Indian alternative to Twitter. We collected a dataset of 4.07 million users, 163.12 million follower-following relationships, and their content and activity across 12 languages. We study the user demographic along the lines of language, location, gender, and profession. The prominent presence of Indian languages in the discourse on Koo indicates the platform's success in promoting regional languages. We observe Koo's follower-following network to be much denser than Twitter's, comprising of closely-knit linguistic communities. An N-gram analysis of posts on Koo shows a #KooVsTwitter rhetoric, revealing the debate comparing the two platforms. Our characterization highlights the dynamics of the multilingual social network and its diverse Indian user base.

Abstract

Recent data-driven approaches to visual servoing have shown improved performances over classical methods due to precise feature matching and depth estimation. Some recent servoing approaches use a model predictive control (MPC) framework which generalise well to novel environments and are capable of incorporating dynamic constraints, but are computationally intractable in real-time, making it difficult to deploy in real-world scenarios. On the contrary, single-step methods optimise greedily and achieve high servoing rates, but lack the benefits of the MPC multi-step ahead formulation. In this paper, we make the best of both worlds and propose a lightweight visual servoing MPC framework which generates optimal control near real-time at a frequency of 10.52 Hz. This work utilises the differential cross-entropy sampling method for quick and effective control generation along with a lightweight neural network, significantly improving the servoing frequency. We also propose a flow depth normalisation layer which amelio- rates the issue of inferior predictions of two view depth from the flow network. We conduct extensive experimentation on the Habitat simulator and show a notable decrease in servoing time in comparison with other approaches that optimise over a time horizon. We achieve the right balance between time and performance for visual servoing in six degrees of freedom (6DoF), while retaining the advantageous MPC formulation. Our code and dataset are publicly available

Abstract

Skin is one of the very important and largest organs of the human body that helps in regulating the body temperature and is responsible for sensations such as feel, touch, hot and cold. These days, skin problems are very common in day-to-day life. There may be many reasons for skin diseases: unbalanced and impure diet, several types of pollutions, or maybe family heredity. However, if skin disease after a long treatment does not show any sign of improvement or the skin cells grow abnormally, this may lead to skin cancer. There are many forms of skin cancer. For early and timely diagnosis of skin cancer, an efficient technique is required at utmost importance. Many people across the globe lost their lives due to the late diagnosis. Therefore, a technique that is cost-effective, quicker, and easily accessible needs a higher demand. These days for the classification of images, machine learning, and deep learning techniques proved to be the most efficient approach. In this paper, the dataset of several images of a benign and malignant tumor was taken and pre-processed. Once all the images were pre-processed, they are ready to fed in several CNN models. These models extract the features and pass the images to several machine learning classifiers for the classification of moles as benign or malignant. The results verify by using the classification approach it becomes very much easy for the dermatologist to easily detect the lesions and provide the appropriate treatment to the patient to save the life.

Abstract

Agriculture is an innovative way of cultivating crops where the resources are optimized to reduce the harmful impacts on environment and to get better crop yields. India has immense opportunities to make use of the potential of Agri-Tech solutions as majority of farmers possess small land holdings. Ecological sustainability is another significant challenge currently needed to be addressed. The paper propose a Smart-Agri architecture based on digital systems with available technologies for incorporating the factors required for better crop yield and to maintain a sustainable ecosystem. The architecture specifically caters to the need of Small Cardamom which is grown in hilly regions and is particularly much delicate to the climate changes. Due to the rampant use of fertilizers and pesticides, lot of degradation has happened to the Cardamom hill reserve zones over past few decades. In the system, data is to be collected from different parts of a farm using mesh connected ZigBee network making a low-power framework, to monitor soil and plant health and analyze data for finding implications to sort out recommendations meeting sustainability and crop goals. The proposed architecture aims to optimize the resources for small farms by assessing the needs of farmlands. This helps to mitigate the adverse effects on nature.

Abstract

In the traditional agriculture in India, the farmers are not able to make profit from their crops. We propose a price prediction model using Machine Learning approaches, which helps the farmers to make appropriate decisions well before cultivation and later on selling their farm outputs. We have chosen cardamom as a case study in which the data being collected over four decades’ monthly average prices. We propose an ensemble method of multivariate linear regression model and ARIMA model over the clustered average monthly cardamom prices across forty years. The robustness of the proposed model is evaluated against four- decade monthly price movemen