arXiv subjects

Himani Arora

Publications and source records attributed to Himani Arora.

8 recordsLinked to original sources

arXiv2026

What Could the Agent See at 19:05? Generating Temporal Enterprise Scenarios from Real Research and Replaying Them to Evaluate Agents

Enterprise AI agents act across many apps whose data changes continuously, so an answer is correct only relative to what data existed and who could see it at the moment it was asked. Offline evaluation today grades against a single static snapshot, effectively the end of the episode. So, it can only evaluate one situation, the final one, even though every earlier moment of the episode is a different situation that invites its own realistic questions with its own correct answers. Recreating each of those moments as a separate snapshot would mean re-provisioning a whole tenant per instant, which is prohibitively costly; and even a single snapshot leaks future state hidden inside records and cannot represent the multi-app, time-ordered way real work happens. Our system closes two gaps at once: it generates a realistic, persona-driven, temporally-evolving enterprise world from real research, and replays that world at any chosen moment to evaluate any pluggable agent. A schema-inferred temporal description drives a deterministic-plus-LLM rebuild of each record's past state; because the queryable moments are finite, all rebuilds are precomputed into a compact difference cache, making evaluation a fast, reproducible lookup with no model in the path. We describe the design, an architecture spanning both flows, and early experience evaluating enterprise agents.

cs.SE↗

arXiv2022

Electrochemical investigation of MoSeTe as an anode for sodium-ion batteries

Sodium ion batteries (SIBs) are considered as an efficient alternative for lithium-ion batteries (LIBs) owing to the natural abundance and low cost of sodium than lithium. In this context, the anode materials play a vital role in rechargeable batteries to acquire high energy and power density. In order to demonstrate transition metal dichalcogenide (TMD) as potential anode materials, we have synthesized MoSeTe sample by conventional flux method, and the structure and morphology are characterized using x-ray diffraction (XRD), field-emission scanning electron microscopy (FESEM), transmission electron microscopy (TEM), and Raman spectroscopy. These characterisations confirm the hexagonal crystal symmetry with p63/mmc space group and layered morphology of MoSeTe. We investigate the electrochemical performance of a MoSeTe as a negative electrode (anode) for SIBs in the working potential range of 0.01 to 3.0~V. In a half-cell configuration, the MoSeTe as an anode and Na metal as counter/reference electrode exhibits significant initial specific discharge capacities of around 475 and 355 mAhg$^{-1}$ at current densities of 50 and 100 mAg$^{-1}$, respectively. However, the capacity degraded significantly like $\approx$200~mAhg$^{-1}$ in 2nd cycle, but having $\approx$100\% Coulombic efficiency, which suggest for further modification in this material to improve its stability. The cyclic voltammetry (CV) study reveals the reversibility of the material after 1st cycle, resulting no change in the initial peak positions. The electrochemical impedance spectroscopy (EIS) measurements affirms the smaller charge transfer resistance of fresh cells than the cells after 10th cycle. Moreover, the extracted diffusion coefficient is found to be of the order of 10$^{-14}$ cm$^2$s$^{-1}$.

cond-mat.mtrl-sci↗

arXiv2022

Terahertz control of photoluminescence emission in few-layer InSe

A promising route for the development of opto-elelctronic technology is to use terahertz radiation to modulate the optical properties of semiconductors. Here we demonstrate the dynamical control of photoluminescence (PL) emission in few-layer InSe using picosecond terahertz pulses. We observe a strong PL quenching (up to 50%) after the arrival of the terahertz pulse followed by a reversible recovery of the emission on the time scale of 50ps at T =10K. Microscopic calculations reveal that the origin of the photoluminescence quenching is the terahertz absorption by photo-excited carriers: this leads to a heating of the carriers and a broadening of their distribution, which reduces the probability of bimolecular electron-hole recombination and, therefore, the luminescence. By numerically evaluating the Boltzmann equation, we are able to clarify the individual roles of optical and acoustic phonons in the subsequent cooling process. The same PL quenchingmechanismis expected in other van derWaals semiconductors and the effectwill be particularly strong for materials with low carrier masses and long carrier relaxation time, which is the case for InSe. This work gives a solid background for the development of opto-electronic applications based on InSe, such as THz detectors and optical modulators.

cond-mat.mtrl-sci↗

arXiv2020

Autocorrected Off-axis Holography of 2D Materials

The reduced dimensionality in two-dimensional materials leads a wealth of unusual properties, which are currently explored for both fundamental and applied sciences. In order to study the crystal structure, edge states, the formation of defects and grain boundaries, or the impact of adsorbates, high resolution microscopy techniques are indispensible. Here we report on the development of an electron holography (EH) transmission electron microscopy (TEM) technique, which facilitates high spatial resolution by an automatic correction of geometric aberrations. Distinguished features of EH beyond conventional TEM imaging are the gap-free spatial information signal transfer and higher dose efficiency for certain spatial frequency bands as well as direct access to the projected electrostatic potential of the 2D material. We demonstrate these features at the example of h-BN, at which we measure the electrostatic potential as a function of layer number down to the monolayer limit and obtain evidence for a systematic increase of the potential at the zig-zag edges.

cond-mat.mes-hall↗

arXiv2020

Photoluminescence dynamics in few-layer InSe

We study the optical properties of thin flakes of InSe encapsulated in hBN. More specifically, we investigate the photoluminescence (PL) emission and its dependence on sample thickness and temperature. Through the analysis of the PL lineshape, we discuss the relative weights of the exciton and electron-hole contributions. Thereafter we investigate the PL dynamics. Two contributions are distinguishable at low temperature: direct bandgap electron-hole and defect-assisted recombination. The two recombination processes have lifetime of $τ_1 \sim 8\;$ns and $τ_2 \sim 100\;$ns, respectively. The relative weights of the direct bandgap and defect-assisted contributions show a strong layer dependence due to the direct-to-indirect bandgap crossover. Electron-hole PL lifetime is limited by population transfer to lower-energy states and no dependence on the number of layers was observed. The lifetime of the defect-assisted recombination gets longer for thinner samples. Finally, we show that the PL lifetime decreases at high temperatures as a consequence of more efficient non-radiative recombinations.

cond-mat.mes-hall↗

arXiv2018

Multi-task Learning for Continuous Control

Reliable and effective multi-task learning is a prerequisite for the development of robotic agents that can quickly learn to accomplish related, everyday tasks. However, in the reinforcement learning domain, multi-task learning has not exhibited the same level of success as in other domains, such as computer vision. In addition, most reinforcement learning research on multi-task learning has been focused on discrete action spaces, which are not used for robotic control in the real-world. In this work, we apply multi-task learning methods to continuous action spaces and benchmark their performance on a series of simulated continuous control tasks. Most notably, we show that multi-task learning outperforms our baselines and alternative knowledge sharing methods.

cs.LG↗

arXiv2017

Lip2AudSpec: Speech reconstruction from silent lip movements video

In this study, we propose a deep neural network for reconstructing intelligible speech from silent lip movement videos. We use auditory spectrogram as spectral representation of speech and its corresponding sound generation method resulting in a more natural sounding reconstructed speech. Our proposed network consists of an autoencoder to extract bottleneck features from the auditory spectrogram which is then used as target to our main lip reading network comprising of CNN, LSTM and fully connected layers. Our experiments show that the autoencoder is able to reconstruct the original auditory spectrogram with a 98% correlation and also improves the quality of reconstructed speech from the main lip reading network. Our model, trained jointly on different speakers is able to extract individual speaker characteristics and gives promising results of reconstructing intelligible speech with superior word recognition accuracy.

cs.CV↗

arXiv2017

Computing Egomotion with Local Loop Closures for Egocentric Videos

Finding the camera pose is an important step in many egocentric video applications. It has been widely reported that, state of the art SLAM algorithms fail on egocentric videos. In this paper, we propose a robust method for camera pose estimation, designed specifically for egocentric videos. In an egocentric video, the camera views the same scene point multiple times as the wearer's head sweeps back and forth. We use this specific motion profile to perform short loop closures aligned with wearer's footsteps. For egocentric videos, depth estimation is usually noisy. In an important departure, we use 2D computations for rotation averaging which do not rely upon depth estimates. The two modification results in much more stable algorithm as is evident from our experiments on various egocentric video datasets for different egocentric applications. The proposed algorithm resolves a long standing problem in egocentric vision and unlocks new usage scenarios for future applications.

cs.CV↗