microstrategy - 2024
Macro
Meso
Micro
learned, data-driven regularizer
Macro
Meso
Micro
tumor
non tumor
false positives
false negatives
necrotic tumor
viable tumor
non-tumor
Manual coarse annotation
Computational refinement
Ground truth
Macro
Meso
Micro
GluA2 AMPAR receptor subunit with a pH-dependent fluorescent tag (SEP) enables in vivo visualization of endogenous GluA2-containing synapses.
GluA2 AMPAR receptor subunit with a pH-dependent fluorescent tag (SEP) enables in vivo visualization of endogenous GluA2-containing synapses.
For an observation \(y\)
\[y = x + \epsilon,~\epsilon \sim \mathcal{N}(0, \sigma^2\mathbb{I})\]
reconstruct \(x\) with
\[\hat{x} = F(y) \sim \mathcal{Q}_y \approx p(x \mid y)\]
\(x\)
\(y\)
\(F(y)\)
Lemma
\(\mathcal I(y)\) provides entrywise coverage for pixel \(j\), i.e.
\[\mathbb{P}\left[\text{next sample}_j \in \mathcal{I}(y)_j\right] \geq 1 - \alpha\]
If \[\mathcal{I}(y)_j = \left[ \frac{\lfloor(m+1)Q_{\alpha/2}(y_j)\rfloor}{m} , \frac{\lceil(m+1)Q_{1-\alpha/2}(y_j)\rceil}{m}\right]\]
\(0\)
\(1\)
low: \( l(y) \)
\(\mathcal{I}(y)\)
up: \( u(y) \)
(distribution free)
\(x\)
\(y\)
lower
upper
intervals
\(|\mathcal I(y)_j|\)
\(0\)
\(1\)
ground-truth is
contained
\(\mathcal{I}(y_j)\)
\(x_j\)
Procedure For pixel \(j\)
\[\mathcal{I}_{\lambda}(y)_j = [\text{lower} - \lambda, \text{upper} + \lambda]\]
choose
\[\hat{\lambda} = \inf\{\lambda \in \mathbb{R}:~\forall \lambda' \geq \lambda,~\text{risk}(\lambda') \leq \epsilon\}\]
ground-truth is
contained
\(0\)
\(1\)
\(\mathcal{I}(y_j)\)
\(\lambda\)
\(x_j\)
Definition For risk level \(\epsilon\), failure probability \(\delta\), \(\mathcal{I}(y_j) \) is a RCPS if
\[\mathbb{P}\left[\mathbb{E}\left[\text{fraction of pixels not in intervals}\right] \leq \epsilon\right] \geq 1 - \delta\]
scalar \(\lambda \in \mathbb{R}\)
\(\mathcal{I}_{\lambda}(y)_j = [\text{low} - \lambda, \text{up} + \lambda]\)
\(\rightarrow\)
vector \(\bm{\lambda} \in \mathbb{R}^d\)
\(\rightarrow\)
\(\mathcal{I}_{\bm{\lambda}}(y)_j = [\text{low} - \lambda_j, \text{up} + \lambda_j]\)
Guarantee: \(\mathcal{I}_{\bm{\lambda}}(y)_j = [\text{low} - \lambda_j, \text{up} + \lambda_j]\) are RCPS
For a \(K\)-partition of the pixels \(M \in \{0, 1\}^{d \times K}\)
\(K=4\)
\(K=8\)
\(K=32\)
\(\hat{\lambda}_K\)
conformalized uncertainty maps
\(K=4\)
\(K=8\)
Teneggi, J., Tivnan, M., Stayman, W., & Sulam, J. (2023, July). How to trust your diffusion model: A convex optimization approach to conformal risk control. In International Conference on Machine Learning. PMLR.
\[\mathbb{P}\left[\mathbb{E}\left[\text{fraction of pixels not in intervals}\right] \leq \epsilon\right] \geq 1 - \delta\]
ground truth pixels!
true whatever your model, whatever your training data,
distribution free
uncertainty intervals:
K-RCPS: High dimensional conformal risk control
What parts of the image are important for this prediction?
What are the subsets of the input so that
Fast hierarchical games for image explanations, Teneggi, Luster & Sulam,
IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022
Teneggi, J., Bharti, B., Romano, Y., & Sulam, J. (2023). SHAP-XRT: The Shapley Value Meets Conditional Independence Testing. Transactions on Machine Learning Research.
Text
\(H_0:\) "The distribution of \(f(x)\) w.r.t. a feature set \(S\) remains unchanged when the \(i^{th}\) feature is introduced"
Teneggi, J., Yi, P. H., & Sulam, J. (2023). Examination-level supervision for deep learning–based intracranial hemorrhage detection at head CT. Radiology: Artificial Intelligence, e230159.
Is the model fair?
Pneumonia
Clear
Pneumonia
Clear
Does your model achieve a \(\Delta_{\text{TPR}}\) of at most (say) 6% ?
Pneumonia
Clear
Tight upper bounds to fairness violations
(optimally) Actionable
Bharti, B., Yi, P., & Sulam, J. (2023). Estimating and Controlling for Equalized Odds via Sensitive Attribute Predictors. Neurips.
Collaborators
Adam Charles (BME)
Peter van Zijl (Radiology)
Xu Li (Radiology)
Paul H. Yi (UMB, Radiology)
Web Stayman (BME)
Mat Tivnan (BME)
Sasha Popel (BME)
Tiger Xu (Neuro)
Dwight Bergles (Neuro)
Rick Huganir (Neuro)
Agustin Graves (Neuro, BME)
Gabrielle Coste (Neuro)
Aaron W. James (Pathology)
Carla Saoud (Pathology)
Sintawat Wangsiricharoen (Patho)
Christa L LiBrizzi (Patho)
Carol D Morris (Patho)
Adam S Levin (Patho)
Game Changers: Artificial Intelligence Part I, 2028