Indian Institute of Science Bangalore
etd@IISc Electronic Theses and Dissertations at Indian Institute of ScienceNot a member yet
6204 research outputs found
Sort by
Methods for Improving Data-efficiency and Trustworthiness using Natural Language Supervision
No full textTraditional strategies to build machine learning based classification systems employ discrete labels as targets. This limits the usefulness of such systems in two ways. First, the generalizability of these systems is limited to labels present and well represented in the training data. Second, with increasingly larger neural network models gaining acceptability, supervision with discrete labels alone does not lead to a straightforward interface for generating explanations for the decisions taken by such systems. Natural Language (NL) Supervision (NLS), in the form of task descriptions, examples, label descriptions and explanations for labelling decisions, provides a way to overcome these bottlenecks. Working in this paradigm, we propose novel methods for improving data-efficiency and trustworthiness:
(1) [Data Efficiency using NLS] Word Sense Disambiguation (WSD) using Sense Definition Embeddings:
WSD, a long-standing open problem in Natural Language Processing (NLP), typically presents itself with small training corpora with long tails of label distributions. Existing supervised methods didn’t generalize well to rare or unseen classes while NL supervision based systems did worse on overall (standard) evaluation benchmarks. We propose Extended WSD Incorporating Sense Embeddings (EWISE), a supervised model to perform WSD by predicting over a continuous sense embedding space as opposed to a discrete label space. This allows EWISE to generalize over both seen and unseen senses, thus achieving generalized zero-shot learning. To obtain target sense embeddings, EWISE utilizes NL sense definitions along with external knowledge in WordNet relations. EWISE achieved new state-of-the-art WSD performance at the time of publication, specifically by improving on zero-shot and few-shot learning.
(2) [Trustworthiness using NLS] Natural Language Inference (NLI) with Faithful NL Explanations:
Generated NL explanations are expected to be faithful, i.e., they should correlate well with the model’s internal decision making. In this work, we focus on the task of NLI and address the following question: can we build NLI systems which produce labels with high accuracy, while also generating faithful explanations of its decisions? We propose Natural-language Inference over Label-specific Explanations (NILE), a novel NLI method which utilizes auto-generated label-specific NL explanations to produce a label along with its faithful explanation. Our evaluation of NILE also supports the claim that accurate systems capable of providing testable explanations of their decisions can be designed.
(3) [Improving the NLS interface of Large Language Models (LLM)]
LLMs, pre-trained on unsupervised corpora, have proven to be successful as zero-shot and few-shot learners on downstream tasks using only a textual interface. This enables a promising NLS interface. A typical usage involves augmenting an input example along with some priming text comprising of task descriptions and training examples and processing the output probabilities to make predictions. In this work, we further explore priming-based few-shot learning and make the following contributions:
(a) Reordering Examples Helps during Priming-based Few-Shot Learning: We show that presenting training examples in the right order is key for generalization. We introduce PERO (Prompting with Examples in the Right Order), where we formulate few-shot learning as search over the set of permutations of the training examples. We demonstrate the effectiveness of the proposed method on the tasks of sentiment classification, natural language inference and fact retrieval. We show that PERO can learn to generalize efficiently using as few as 10 examples, in contrast to existing approaches.
(b) Answer-level Calibration (ALC) Helps Free-form Multiple Choice Question Answering (QA): We consider the QA format, where we need to choose from a set of free-form textual choices of unspecified lengths, given a context. We present ALC, where our main suggestion is to model context-independent biases in terms of the probability of a choice without the associated context and to subsequently remove these biases using an unsupervised estimate of similarity with the full context. ALC improves zero-shot and few-shot performance on several benchmarks while also providing a more reliable estimate of performance
Investigation of anticancer compounds in the natural productome of marine algae associated-endophytic fungi
No full textCancer is one of the leading causes of death worldwide. Owing to the complex ways in which this disease develops, there is a constant demand for new drugs. Natural products have delivered great promise in inspiring chemotherapy, for example, Taxol, Trabectedin, Plinabulin, Marizomib, Midostaurin, etc. Endophytic fungi have potentiated this promise by producing potent bioactive compounds. They live symbiotically in the internal tissues of higher organisms. Marine endophytic fungi offer a diverse chemical space for the discovery of novel anticancer compounds, sustainably. Marine endophytes demonstrate a higher chemical diversity, possibly due to immense competition and abiotic and biotic stress. These secondary metabolites are highly potent at minuscule concentrations because they act in dilute environments over larger distances.
In this study, 26 endophytic fungi were isolated and identified from 10 marine algal samples collected from the Konkan coast, India. This work led to the first diversity study of marine algae-associated endophytic fungi of the Konkan coast. All the fungal extracts were screened for cytotoxicity on several human cancer cell lines, A431, HeLa, A549 and MCH-7. Three endophytic fungi, Aspergillus unguis AG 1.1 (G), Chaetomium globosum PG 1.6, and Aspergillus unguis AG 1.2, were chosen for further study as their IC50 values were < 10 μg/mL.
Chrysin, a dihydroxyflavone was purified and reported for the first time from a marine endophyte C. globosum. It was so far known to be found in honey and passionflower. It induced apoptosis, G1 phase cell cycle arrest, MMP loss, and ROS production in MCF-7 cells. Further, using metabolic profiling of C. globosum, several key intermediates of the chrysin biosynthesis pathway were identified, thus, proving the presence of flavonoid biosynthetic machinery in the marine fungus. The yield of chrysin was enhanced using optimization of several media and fermentation parameters, inducing abiotic and biotic stresses and elicitation. The free use of chrysin in clinical scenarios is disadvantaged due to poor solubility at physiological pH, rapid metabolism, and low bioavailability. In this study, chrysin nanoparticles (NChr) were prepared by optimization of several physico-chemical parameters using response surface methodology. The apoptotic effect of NChr was studied in HeLa cells by Attenuated Total Reflectance-Fourier Transform Infrared Spectroscopy.
This work also reports the isolation and characterization of a novel isoquinoline derivative from A. unguis AG 1.1 (G). It had a N-heterocyclic steroid-like structure with a lactone moiety. The compound demonstrated maximum cytotoxicity in A549 cells with an IC50 of 18.8 μM after 48 h. It induced G1 phase cell cycle arrest, disruption of mitochondrial membrane potential, production of reactive oxygen species and caspase 3/7 activation, leading to apoptosis.
This work highlights the immense potential of marine natural products that can introduce highly impactful drugs into the pharmaceutical market
PLGA Based Drug Carriers for Treatment of Osteoarthritis
No full textOsteoarthritis (OA) has affected nearly 22-39% of the Indian population and remains one of the most common musculoskeletal disorders that affects articular joints. Post-traumatic osteoarthritis, which develops following trauma to the knee joint, can lead to severe disability and morbidity in otherwise healthy individuals. The current standard of care for osteoarthritis (OA) primarily revolves around symptomatic relief with total knee arthroplasty at the end stage of the disease. Translation of many disease-modifying OA drugs (DMOADs) stagnates due to low bioavailability at the affected site, necessitating multiple injections and decreased patient compliance. Our project's primary aim was to encapsulate DMOADs in a polymer matrix to assess their tuneability and evaluate the drugs' potency in preventing and treating OA.
Potential disease-modifying strategies widely researched are the modulation of autophagy and senescence pathways. Rapamycin and Nordihydroguaiaretic acid (NDGA) hold promise as a potent DMOAD in OA as they induce autophagy and prevent senescence. NDGA is also a potent ROS scavenger and can reduce oxidative stress in OA. Although intra-articular injections of these drugs can ensure delivery of large doses in the knee joint, lymphatic clearance of drug can still rapidly reduce the drug concentration below the therapeutic levels making the translation of these drugs difficult.
Here, we have synthesized drug-loaded poly (lactic-co-glycolic acid) microparticles (MPs) that induced autophagy, prevented senescence, and sustained sGAG production in primary human articular chondrocytes from OA patients. These MPs were potent, nontoxic, and exhibited prolonged retention time (up to 35 days) in mice joints. Intra-articular delivery of these MPs effectively mitigated cartilage damage and inflammation in surgery-induced OA when administered as a prophylactic or therapeutic regimen in the regular diet and high fat-fed obese mice models. Together, our studies demonstrate the feasibility of using rapamycin-loaded and NDGA-loaded MPs as potential clinically translatable therapies to prevent and treat post-traumatic osteoarthritis
Multi-connectivity for Urllc and Coexistence with Embb in Time-varying and Fading Channels
No full textUltra-reliable and low latency communications (URLLC) is a novel use case of 5G. It has challenging requirements like such as low block error rates (BLERs) and stringent latency targets. Multi-connectivity, in which multiple base stations (BSs) transmit the same data to the user, is a key technique in 5G to meet the stringent reliability requirements. URLLC data is immediately transmitted by puncturing the ongoing enhanced mobile broadband (eMBB) transmission to satisfy the latency constraint. However, this results in an increase in the BLER of the eMBB users.
We first propose a low complexity multi-connectivity MCS selection algorithm (MCMSA) to select the subset of co-operating BSs and the modulation and coding schemes (MCSs) they employ. The goal is to minimize the eMBB throughput loss while satisfying the URLLC constraints. We study two types of multi-connectivity: orthogonal transmission (OT) and joint transmission (JT). We derive tractable expressions to calculate the achievability, which is the probability that the URLLC reliability requirement can be satisfied by the multi-connectivity. We do so for flat fading and frequency-selective fading scenarios. Our results highlight the trade-offs between URLLC achievability, eMBB throughput loss, and channel state information (CSI) feedback overhead of OT and JT.
We then consider time-varying channels, in which the CSI report from the URLLC user to the BSs about the downlink channel gains becomes partially outdated by the time the BSs transmit data. We propose a novel stochastic BLER constraint for selecting MCSs at the time of transmission. We derive expressions for the conditional probability that the BLER of an MCS at the time of transmission is less than the target given the CSI fed back. These expressions enable the selection of MCS at the time transmission and meets the URLLC error target with high probability. Our results bring out the significant impact of feedback delays on reliability even at moderate Doppler spreads
Low delay file transmissions over power constrained quasi-static fading channels
No full textThe ubiquitous deployment of battery-operated wireless devices has resulted in the need
for efficient low latency power allocation schemes. A common phenomenon in wireless
transmission systems is congestion, where the transmitter backlog grows due to restrictions
in channel usage on a resource-constrained shared access medium. In this research
work, we aim to achieve low communication delay of wireless downlink file transmissions
operating on power-constrained quasi-static fading channels, using state-dependent transmission
rate control and admission of file transmission requests. We employ a Markov
queueing model to formulate the low delay objective for exponentially distributed file sizes
as a constrained average queue length minimization problem.
The corresponding primal problem is known to be expressible as a linear program
in occupation measures, and therefore strong duality holds. In our work, we show the
primal feasibility of the dual optimal policy w.r.t. the average throughput and power constraints,
which is proved under the assumption the optimal average power and throughput
are continuous with respect to the Lagrange dual variables at the optimal point. The dual
problem is simplified to an iterative optimization using Dinkelbach’s fractional programming
method and solved using gradient analysis techniques to analytically derive the
ON-OFF threshold characteristics of the admission policy and the recursive structure of
the transmission rate policy.
We first apply our solution method to a wireless transmission system using the M/M/1
queueing model. Our objective is to minimize the average queue length subject to an upper
bound on average transmission power and a lower bound on average admission rate.
This constrained average queue length minimization problem is solved using Lagrange dual method. We substitute the individual stationary probabilities in the Lagrange dual
function using the product form distribution expressed in terms of the stationary probability
of the maximum queue length. The resulting objective function then corresponds
to a fractional minimization problem which is solved using Dinkelbach’s method. We
analytically derive the ON-OFF threshold characteristic of the optimal admission rates
and the recursive structure of the optimal transmission rates. We illustrate the results of
our algorithm for different values of throughput and power requirements. We also demonstrate
the efficiency of optimal state-dependent rate control for exponentially distributed
file sizes compared to benchmark state-independent transmission schemes.
We next apply the solution techniques to an energy harvesting wireless transmission
system, extending the M/M/1 queueing model. The model uses energy stored in a battery
as well as energy packets available from an auxiliary power supply for file transmission. We
use the product-form stationary distribution to establish a correspondence between the
energy harvesting system and the M/M/1 queueing system. Using the solution approach
using Dinkelbach’s method, we derive similar characteristics for the optimal admission
and transmission rates.
We finally extend the analysis to model a cache-aided wireless transmission system
operating under the assumption the cache-hit probability is uniform for all files and queue
length states. The system is modeled as a quasi-one-dimensional Markov chain. The stationary
probabilities in the Lagrange dual function are expressed in terms of the stationary
probability of the empty buffer state using the product of matrices. The solution methods
and insights developed from the previous models simplify the analysis of this problem,
and we analytically characterize the structure of the optimal admission and transmission
rates. The applicability of our solution methodology to these three models of transmission
systems illustrates its simplicity and versatility
Transport of Intensity Equation based Quantitative Phase Imaging of Red Blood Cells
No full textThis thesis reports the characterization of Red Blood Cell (RBC) morphology in a non-contact and
label-free manner using a flexible, low-cost continuous imaging system. Detection of abnormalities
in red blood cell properties, including shape, size, and number, can reveal a range of pathologies. A
usual laboratory hematologic diagnosis consists of a complete blood count (CBC) and a peripheral
blood smear (PBS) review. An Automated Hematology Analyzer reports the complete blood count,
which includes Hematocrit, Hemoglobin content in RBCs and Total Count, Mean Cell Volume,
Distribution width for each blood cell (RBC, WBC, Platelets). But they fail to provide any
information about the cell shape, which is an essential property in determining cell morphology.
A peripheral blood smear analysis involves imaging RBCs under a microscope and determining
cell shape, although cell volume cannot be found as cell thickness remains unknown. Since cells
are imaged in dry form and distributed non-uniformly, finding cell count using a PBS review is
not possible. So, there is a need for a different technique that can be used to study red blood
cell morphology effectively. Today one of the rapidly growing research fields in studying cell
morphology and cell dynamics is Quantitative Phase Imaging (QPI). It combines advancements
in optics, imaging theory, and computational methods to image phase information of the sample
quantitatively. In this work, determination of RBC total count, MCV, and RDWusing a single, fast,
portable, and cost-effective optical setup has been proposed. It involves quantifying the phase delay
introduced by the red blood cells using a QPI method called the Transport of Intensity Equation
(TIE). The application of TIE as a QPI method does not require complex setups or expensive
components, unlike other QPI techniques. A partially coherent light beam from a conventional
LED is used to illuminate a diluted blood sample loaded in a microfluidic channel. Through-focus
intensity images of RBCs arranged in a monolayer are acquired using a low-cost continuous imaging
system. Intensity images are processed using a Fast Fourier Transform (FFT) based Poisson solver
to find a solution to the transport of intensity equation. The solution to TIE is the phase distribution
of light at the focus. From phase, the thickness profile of each cell can be calculated, hence the cell
volume. RBC total count is calculated by counting the cells in the given field of view. Cell shape
is determined from the focal image. So the proposed system provides complete information about
the red blood cells morphology at a much lower cost than hematology analyzers and peripheral
blood smear analysis. The overall aim of this research project is to explore the potential of our
imaging system combined with the TIE algorithm as a reliable tool for characterizing red blood
cells morpholog
Accelerated Search of Catalysts Using Density Functional Theory and Machine Learning
No full textThe need for clean and renewable energy resources has propelled the interest in designing new catalysts producing energy from renewable resources and alternate cleaner fuels such as hydrogen, methane, ammonia, ethylene, etc. Despite an extensive search, finding an efficient catalyst in terms of activity, selectivity, stability, and cost is still far from reality. We attempt to address some of these challenges by combining the density functional theory (DFT) and machine learning (ML). We report a carbon-nitride and transition metal (TM) based single atom catalyst (TM-SAC) for electrocatalytic nitrogen reduction reaction (eNRR). Among all the TM-based SACs, Mo- and W-SACs are found to be highly active and selective for eNRR over competing hydrogen evolution reaction (HER). The higher activity is attributed to the optimum stability of nitrogen and other eNRR intermediates over the SAC. Further, we addressed one of the major issues of CO2 reduction reaction (CO2RR) catalysts i.e., their selectivity. Our work presented a simple solution where the selectivity can be tuned by varying alloy surface composition. This arises due to the change in electronic structure of the bimetallic catalyst, simultaneously changing the d-band center of the metals. Modifying the surface composition is further employed to enhance the activity of Pt-Pd based alloys for methanol oxidation reaction (MOR) and oxygen reduction reaction (ORR) for fuel-cell applications. Surface Pd helped altering the thermodynamics of the reaction, which is also confirmed by experimental validation. Importantly, the d-band center emerged as the descriptor for the activity of the proposed catalysts. Owing to the complexity and resource extensive calculations involved in determining the catalytic activity of the alloy catalyst, we employed machine learning approach to estimate the d-band center of core-shell nanoparticles, a measure of catalytic performance. The machine learning model based on recommender-system uses data calculated from DFT for d-band center and a collection of accessible elemental information. This model recommends bimetallic nanoparticles with an optimum range of d-band center and capture the pair-wise properties of the metals present in core and shell of the nanoparticles. Further, we developed an efficient framework to search for a water-splitting photocatalyst by using metal phosphorus trichalcogenides (MPX3) class of compounds. The high-throughput study corroborates the role of accurate band gap, band-edge alignment, optical transitions, and charge carrier mobilities of the materials. The thermodynamics of the redox reactions confirms the photocatalytic efficiency of the screened catalysts. The results of our study pave way to overcome some of the critical challenges related to catalysts by effectively addressing the selectivity and activity problems of both existing and newly designed catalysts
Modelling, Stabilization Methods and Power Amplification for Power Hardware-in-Loop Simulation with Improved Accuracy
No full textAccurate testing of a device under development, in near to real-life environment, is essential
to the rapid growth of emerging technologies such as distributed generation, renewable energy
sources, electrical storage, microgrid and electrical vehicles. Power hardware in loop (PHIL)
simulation is an emerging methodology which allows for the testing of a physical hardware,
i.e., a device under test (DUT), in a safe and controlled environment, without the rest
of the system being available. The rest of the system for the DUT is emulated through
its mathematical models computed in a real-time simulator (RTS). The DUT and the RTS
interact with each other through a power amplifier (PA), which scales up the signals provided
by the RTS for the DUT, and sensors.
For accurate testing of a DUT, its response in a PHIL simulation should be similar to that
in a real-life environment. The two responses are often different due to several factors which
are present in a PHIL simulation but not in an actual system. An RTS is a discrete-time
domain system while the DUT is a continuous-time domain system. Finite computation time
requirement of the RTS and conversion of the continuous-time domain signals to the discretetime domain signals, and vice versa, results in an inaccurate response at high frequencies.
Even worse, the PA employed usually introduces additional dynamics into a PHIL simulation
and results in inaccuracies at much lower frequencies. The inaccuracy can be so significant
that the PHIL simulation of a system can be unstable even though the actual system is
stable. This thesis deals with the power amplification required for the accurate replication
of the fast transients of a system in a PHIL simulation, accurate modelling of high frequency
phenomena of the discrete-time domain characteristics of a PHIL simulation and associated
stability analysis, and stabilizing methods for PHIL simulation.
Conventional linear PAs employed in a PHIL simulation are too bulky, expensive and
lossy to be used at medium to high power applications. Whereas, a conventional filter-based
switched-mode PA has a limited dynamic response to emulate fast transients of a system in
a PHIL simulation. An output filter-less voltage source inverter is proposed as a high bandwidth PA for a PHIL simulation with inductive DUT. The proposed PA, realized through
an IGBT-based PWM converter stack, is utilized to emulate the transients of synchronous
generator, including fast transient corresponding to the field excitation controller, while feeding a balanced linear load. Along with a proposed modification to the conventionally used
synchronous generator model, in order to include the effects of stator transients, improved
accuracy is obtained for unbalanced and non-linear loads also. The applicability of the proposed PA is extended for it to be interfaced to PWM converters by proposing an in-phase
synchronization of PWM carriers of the PA and the converter under test. The PA is utilized
for testing the control of a three-phase 415 V, 3 kW PWM rectifier
For various applications, a PA is required to have power sinking capability, which can
be achieved by supplying it from a grid-connected PWM rectifier. A simple input voltage
sensor-less vector control of PWM rectifier is proposed in the thesis. While the performance
of the proposed method, in terms of THD and power factor, is comparable to the sensorbased method and existing sensor-less methods, its computation time requirement is much
lower than those for these methods. The proposed control is validated through simulations
and experiments on a three-phase 415 V, 3 kW grid-connected PWM rectifier, generating
800 V dc supply.
Conventional continuous-time domain transfer functions of current control loop of a PWM
rectifier, and PWM converters in general, do not represent accurately the closed-loop system
when the bandwidth of the loop is comparable to the switching frequency of the converter.
Third-order reference and disturbance transfer functions of discrete-PI controlled current
loop of PWM converters are proposed in the thesis which are utilized to derive closed-form
expressions for the current response of the converter for step changes in current reference
and voltage disturbance. Consequently, optimized PI-controller parameters are obtained for
the fastest disturbance rejection settling time. Further, a pre-filter to the current control
loop is proposed to achieve dead-beat reference tracking response. The pre-filter, along with
the optimized PI-controller, results in reference tracking and disturbance rejection settling
times of two and eight switching cycles, respectively.
A PHIL simulation, being a combinational of continuous-time domain and discrete-time
domain systems, is conventionally represented using continuous-time models. A discretetime domain model is proposed in the thesis which represents a PHIL simulation much
more accurately than the conventional model. The proposed model is used to conduct
stability analysis of PHIL simulations. The stability limits in terms of the parameters of
the simulated and physical quantities are more accurately estimated through the proposed
method as compared to the conventional methods. The stability limits of PHIL simulation
are verified through simulations and experiments.
Low-pass filter (LPF) based feedback current filtering (FCF) method is widely used
for stabilizing an unstable PHIL simulation. The proposed discrete-time domain modelling
method is utilized to show that the low-pass filter-based FCF method is ineffective in stabilizing PHIL simulations having highly inductive physical impedances. Phase-lag compensator
(PLC) is proposed to be a superior alternative to low-pass filter in such cases. Further, a
novel cross-coupled compensator (CCC) is proposed in this thesis. The same is utilized as
a filter in FCF method for stabilizing those PHIL simulations where both LPF and PLC
are ineffective. CCC-based FCF method is employed for the PHIL simulation of a single
machine infinite bus system. The proposed CCC is also utilized for realizing a fault-tolerant
synchronous inverter, i.e., a renewable energy source-fed grid tied-inverter which is controlled
to act as a synchronous generator. The proposed synchronous inverter draws significantly
less current in case of a grid fault as compared to a conventional synchronous inverter and
hence avoids damage to the inverter without additional current limiting methods.DST, MHR
Deep Learning in Computer Vision: Studies in Neuro-image Segmentation and Satellite Image Super-resolution
No full textSingle image super-resolution (SR) has been a topic of great interest in the computer vision
and deep learning community and has found applications in many areas including quality
enhancement of satellite images. As the cost of satellite images primarily depends on the sensor
quality, super-resolving satellite images captured at a low resolution may substantially reduce
the price of image acquisition. However, none of the existing deep CNN based satellite image
super resolution techniques takes region-level context information into account and gives equal
importance to each image region. Satellite images are typically of very large dimensions and
salient object regions often occupy a small portion of the same. This, along with the fact that
most state-of-the-art SR methods are complex and cumbersome deep models, the time taken
to process very large satellite images can be impractically high. These observations motivate
us to propose a context-aware SR pipeline for satellite images. Specifically, in the first work,
We, propose to handle this challenge by designing an SR framework that analyzes the regional
information content on each patch of the low-resolution image and judiciously chooses to use
more computationally complex deep models to super-resolve more structure-rich regions on the
image, while using less resource-intensive non-deep methods on non-salient regions. Through
extensive experiments on a large satellite image, we show substantial decrease in inference
ime while achieving similar performance to that of existing deep SR methods over several
evaluation measures like PSNR, MSE and SSIM. Finally, as a direct improvement above this
work, we propose a switch-guided hybrid network that is trained to selectively super-resolve
salient regions using a deep CNN model and non-salient/background regions via a lightweight
SR method such as bi-cubic interpolation. Through experiments on the SpaceNet dataset, we
study how the proposed switched SR framework can maintain a balance between computational
cost and improvement in image quality
Superconducting qubit-based hybrid devices
No full textQuantum technology has potential applications in many areas of science and engineering. Recently, there has been significant progress in developing hybrid quantum devices using a superconducting qubit platform. The hybrid devices in this category combine the advantage of superconducting qubits with other degrees of freedom. Many experimental realizations of such devices have demonstrated entanglement, state preparation, and readout between multiple modes. I will present a hybrid electromechanical device consisting of a transmon type qubit and a SiN-based mechanical resonator. The device shows a large coupling between electrical and mechanical degrees of freedom compared to the earlier demonstration in traditional optomechanics. The large coupling manifests itself in the form of LZS interference observed in the qubit spectroscopy. In addition, I will present a theoretical model to understand the three-mode hybrid system in the presence of the external drive. In such a tripartite system, we analyze the steady-state occupation of the mechanical mode to show that the sideband cooling of the mechanical mode to its ground state is achievable. The theoretical calculations here predict the experimental parameters for the optimal readout of the mechanical mode, which is also verified experimentally. In the second part of my talk, I will discuss a novel architecture to implement a fast frequency tunable qubit in a three-dimensional waveguide cavity. Control over the qubit frequency can be a valuable resource in the hybrid system consisting of superconducting qubits. We investigate the flux-dependent dynamic range, relaxation from unconfined states, and the bandwidth of the flux-line. We use the fast-flux line to tune the qubit frequency and demonstrate the swap of a single excitation between cavity and qubit mode. The circuit QED setup presented here provides an alternating method to design a modular hybrid system where the components can easily be modified, added, or removed as required for a design