Sambit Mishra

Ming Hsieh Department of ECE · University of Southern California

Sambit Mishra

PhD student in Electrical & Computer Engineering, advised by Prof. Urbashi Mitra.

I work on causal inference and probabilistic graphical models. My research develops identifiability theory and scalable continuous-optimization methods for causal discovery: characterizing when causal structure can be recovered from data, designing efficient algorithms for identifiable directed acyclic graphs, and establishing variance bounds for the non- and partially-identifiable cases that arise with high-dimensional, mixed-distribution data.

View researchCurriculum VitaeGoogle Scholar
X₁X₂ZY₁Y₂Y₃

fig. 01 — a directed acyclic graph

§ 01 — Research

Research interests

Recovering causal structure that is both correct and computable — when it can be identified, how to estimate it efficiently, and how much uncertainty remains when it cannot.

  1. 01
    Causal discovery
    Continuous-optimization formulations of structure learning that replace combinatorial search over DAGs with differentiable acyclicity constraints, enabling GPU-accelerated estimation at scale.
  2. 02
    Optimized interventions
    Soft-intervention selection that maximizes identifiability gain per experiment, reducing the number of interventions needed to recover causal structure.
  3. 03
    Identifiability
    Conditions under which causal structure is recoverable from finite samples and latent variables, with variance bounds for the non- and partially-identifiable regimes.

§ 02 — Publications

Selected publications

View all research →
  1. [01]

    ICASSP 2026

    Barcelona, Spain

    Learning to Intervene: Optimized Soft Intervention Selection for Causal Discovery

    C. Peng, S. Mishra, U. Mitra

    doi: 10.1109/ICASSP55912.2026.11460954
  2. [02]

    IEEE TGCN, vol. 10

    2026

    SER-Optimized Multi-Level ASK Modulations for RIS-Assisted Communications With Energy- and Sign-Based Noncoherent Reception

    S. Mishra, S. P. Dash, G. C. Alexandropoulos

    doi: 10.1109/TGCN.2025.3633182
  3. [03]

    IEEE WCL, vol. 15

    2026

    Error Analysis With Optimal Receiver and Multi-Level ASK for RIS-Assisted Noncoherent Wireless System

    S. Mishra, S. P. Dash

    doi: 10.1109/LWC.2025.3624154

§ 03 — Background

Background

Before USC, I studied Electronics & Communication Engineering at IIT Bhubaneswar, graduating with a B.Tech (Hons.) in 2025 (9.14/10.00 GPA). There I worked with Dr. Soumya P. Dash on SER-optimized modulation schemes for RIS-assisted noncoherent wireless systems, published in IEEE TGCN and IEEE WCL.

In the summer of 2024 I was an ASIC Engineering Intern at NVIDIA, verifying CHI protocol compliance on the CHI-VIP team. I also served as General Secretary of the Science & Technology Council at IIT Bhubaneswar, leading the institute’s Inter-IIT Tech Meet contingent to a top-ten finish among 23 IITs.

My current work sits at the intersection of optimization and inference: when causal structure can be recovered efficiently, and what guarantees can be placed on the result.