Venue Map

1 Why This Page Matters

This page is not a prestige chart.

It is a fit guide for readers who want to understand where different kinds of papers usually belong, and what sort of evidence each venue family expects.

The practical question is:

if I make this kind of claim, what audience is likely to care, and what proof or experiment culture will they expect?

2 Read Venue Families, Not Rankings

Single venue names are often misleading when read in isolation.

It is usually more useful to think in families:

  • broad ML venues
  • learning-theory and statistics venues
  • optimization and numerical venues
  • control and systems venues
  • signal and communication venues

Each family has a different balance of theorem, experiment, system detail, and application framing.

3 A Practical Venue Map

Venue family Best fit What usually carries the paper Common weak pattern Official entry points
Broad ML methods, models, representation learning, theory-plus-experiment work a coherent empirical story, strong baselines, and enough theory to clarify the mechanism when theory matters paper sounds theoretically deep but experiments do not test the stated claim NeurIPS, ICML, ICLR
Learning theory and statistics sample complexity, generalization, estimation, concentration, asymptotics theorem quality, assumption clarity, proof insight, and carefully scoped experiments if experiments appear at all decorative experiments or theorems whose assumptions do not match the motivating problem COLT, AISTATS, JMLR, TMLR
Optimization-heavy ML and inference work solver behavior, convergence, certificates, conditioning, structured recovery formulation clarity, guarantees tied to algorithm design, and computational evidence that matches the theory large benchmark tables used as a substitute for explaining the mathematical object being solved NeurIPS, ICML, AISTATS
Control and systems dynamics, feedback, estimation, planning, stability, sequential decision-making state-space clarity, stability or performance guarantees, and experiments or simulations that respect system constraints applying ML framing without making the controlled system or performance objective precise IEEE Control Systems Society, Conference information, CDC 2026
Signal and communication filtering, coding, detection, reconstruction, sensing, inverse problems noise model clarity, reconstruction or error metrics, and careful handling of bandwidth, rate, or sensing constraints mixing decoding, denoising, and reconstruction claims without saying which objective is primary ICASSP

4 How To Use This Map

When reading or shaping a paper, ask these questions in order:

  1. What is the main claim type? theorem, algorithm, benchmark, system, or application
  2. What evidence actually supports that claim? proof, simulation, ablation, benchmark, or some mixture
  3. Who is the natural audience? broad ML, theory, optimization, control, signal, or statistics
  4. Which venue family already expects that evidence culture?

If those four answers disagree, the paper often feels confused even before any reviewer points out a flaw.

5 How This Connects To The Site

Back to top