Due to resource constraints or external and internal uncertainties, clients in real-world federated learning systems are often intermittently available edge devices. In highly dynamic environments, the parameter server lacks prior real-time knowledge of clients' availability, making it challenging to adapt traditional federated learning algorithms to be resilient to uncertainties in client availability. If not carefully addressed, complex client availability can introduce significant bias, potentially harming the performance of the trained model. Most prior work either fails to account for non-stationary client availability dynamics or demands significant memory and computational overhead. This paper aims to develop efficient federated learning algorithms that are provably resilient to heterogeneous and non-stationary stochastic client availability. We propose FedSWE, which admits novel algorithmic structures to (i) compensate for missed computations, (ii) stabilize and diffuse the global updates over rounds, and (iii) evenly mix the local updates through implicit gossiping, despite being agnostic to non-stationary dynamics. Compared with the standard FedAvg, FedSWE introduces light additional memory and computation overhead. We show that FedSWE converges to a stationary point of non-convex objectives while achieving the desired linear speedup property in certain special cases. We corroborate our analysis with numerical experiments over diversified client unavailability dynamics on real-world data sets.
Ming Xiang, Stratis Ioannidis, Edmund Yeh et al.· 0 citations
The Duckworth-Lewis-Stern (DLS) method has been the international standard for revising target scores in rain-interrupted limited-overs cricket since 1999. Despite over two decades of operational use, no large-scale empirical audit of its prediction bias has been published. We conduct such an audit on 8,150 international matches (3,095 ODIs, 5,055 T20Is) from Cricsheet, generating 233,550 synthetic interruption scenarios with temporal splits. We document two structured biases. First, DLS prediction error spans a 137-run range across (overs-remaining, wickets-lost) match-state buckets. Second, DLS exhibits a gender-differential bias on ODIs that has not previously been quantified: on the training split, mean over-prediction is +1.51 runs for men but +7.63 runs for women, a gap of +6.13 runs (F = 195.16, p < 10^-43). We benchmark DLS against five modern alternatives: Bi-LSTM, XGBoost, an enriched XGBoost variant, a deep context-aware model, and a stacking ensemble, and propose DLS-Cal, a lightweight interpretable calibration layer (27K parameters) outputting a state-conditioned correction added to DLS. DLS-Cal reduces absolute bias by 31% on ODI and 19% on T20I, and a gender-aware variant reduces women's ODI residual bias from +6.19 to +0.65 runs while leaving men's calibration unchanged. We release code, models, and data.
Designing viable drug candidates requires searching a combinatorially large and rugged chemical space for molecules that satisfy multiple, often competing, objectives. Large language models (LLMs) provide a useful generative prior for this problem because of their representational capacity, reasoning ability, and flexibility when incorporating information from the external environment. While reinforcement learning from verifiable rewards (RLVR) can be used to improve the capabilities of LLMs, many chemically relevant scoring functions require hours or even days per evaluation, making them prohibitively expensive to use directly during online training. Here, we investigate whether LLMs can learn molecular design strategies from cheaper synthetic tasks that generalize to expensive molecular lead optimization settings. We find that curriculum-based training recipes that gradually incorporate more challenging synthetic design tasks enable strong performance that surpasses that of much larger frontier models on structure-based lead optimization. Our results suggest that scaling post-training using synthetic tasks is an effective strategy for adapting LLMs to high-cost experimental scenarios that are too expensive to directly train on.
Frank Hu, Shriram Chennakesavalu, Zichen Wang et al.· 0 citations
Where inside a language model does refusal live, and does that place change when the architecture does? In a transformer, refusal is governed by a single direction in the residual stream, a finding that safety and interpretability tooling now depend on. State-space models (SSMs) route information through a recurrent update instead of attention, sharing no token-mixing mechanism with a transformer. Does the same safety representation survive this shift, or must it be rediscovered per architecture? It survives. A single rigid rotation, which can only reorient a space and not reshape it, aligns one model's representation space with another's, so the two genuinely share the representation. A harm probe trained on a transformer then flags an SSM's harmful inputs, and removing the aligned direction makes a model answer attacks it would otherwise refuse, while a random direction of the same size does far less. What is architecture-specific is not where the direction is steered but where it must be read. Each layer computes a fresh output that is then added into the residual stream, and harm is cleanly readable at this output, the write site, before the addition. A control that holds the intervention's strength fixed shows that what matters is where the direction is estimated, not where it is applied. Applied through a detector-triggered gate, this direction lowers jailbreak success in all four architecture families we test (SSM, transformer, recurrent, hybrid), and on the SSM it holds against an attacker that tunes its prompt against the defense. The gate only matches a trivial rule that returns a fixed refusal whenever the same detector fires, so what transfers across architectures is the direction itself, not defense strength. Safety tooling built on refusal therefore ports to a new architecture by re-estimating the direction at that architecture's write site, not by rebuilding it.
We introduce WEECFP, a parameter-free 1024-dimensional continuous molecular fingerprint that scatters each Morgan substructure across roughly thirty-two signed positions of a single vector, and WEECFP-SuRGE, a transformer architecture whose self-attention applies SuRGE (Substructure Rotary Graph-distance Encoding) -- a RoPE-like rotation parameterized by molecular shortest-path graph distance -- to WEECFP substructure tokens. A 7-model blend of this architecture (the WEECFP-SuRGE Blend) achieves the lowest average regression rank on the TDC ADMET leaderboard; is #2 overall on the TDC ADMET leaderboard (behind only pretrained MapLight+GNN), and is #1 overall among methods that use no external pretraining; takes leaderboard #1 finishes on Pgp, Lipophilicity, CYP2D6 Substrate, Clearance Microsome, and LD50 (with the WEECFP-NoSuRGE Blend separately reaching #1 on HIA) across the full 22-benchmark suite -- without any external pretraining. On MoleculeNet, WEECFP-SuRGE beats every classical-fingerprint baseline on 3 of 4 regression tasks (ESOL, Lipophilicity, QM9). We further show that WEECFP tokenization is near-lossless: a greedy overlap reconstruction recovers the exact canonical SMILES of 99.9% of in-distribution molecules across 9 MoleculeNet datasets and 98.93% of molecules in a cross-dataset holdout (HIV->Lipophilicity), and that a three-reference farthest-first encoding of graph distance correlates at Pearson r = 0.901 with the true pairwise distance, enabling O(S) positional memory at matching accuracy.
We show that susceptibilities, an interpretability technique developed for neural networks, can identify the presence of algorithmic structure in Turing machines by probing the local loss landscape of a learning problem for noisy Turing machines introduced by Murfet and Troiani (arXiv:2504.08075). We prove that symmetries and path separation in the algorithm implemented by a Turing machine induce permutation symmetries and low-rank blocks in its susceptibility matrix. We study this empirically on a set of deterministic finite automata (DFAs) and demonstrate that algorithmic features can be recovered by principal component analysis and clustering methods in susceptibility space.
Billy Snikkers, Rumi Salazar, Daniel Murfet et al.· 0 citations
Symbolic regression (SR) discovers closed-form mathematical expressions from data, offering interpretability beyond black-box models. Existing methods suffer from slow convergence in combinatorial search spaces and lack mechanisms to exploit compositional structure in the data. We introduce SMILE (Sine, Multiplication, Identity, Logarithm, Exponential), a hybrid framework that unifies continuous gradient-based optimization with discrete symbolic recovery through three stages: structural analysis of the data to identify the compositional hierarchy of the target expression, continuous optimization to learn parameters of a network that encodes the target expression using interpretable activations, and symbolic recovery through structured pruning, coefficient optimization, and rounding. This final stage distills the learned network into a compact expression with exact symbolic constants. We evaluate SMILE on SRBench across ground-truth and black-box datasets, with ablation studies validating each component. SMILE achieves the highest symbolic solution rate at the largest noise levels, demonstrating strong robustness where competing methods degrade substantially. It consistently lies on the Pareto front of accuracy versus complexity, recovering significantly simpler expressions in a fraction of the time required by the competing methods.
Mansooreh Montazerin, Antonio Ortega, Ajitesh Srivastava· 0 citations
Neural ocean emulators are being proposed for regional forecasting in cyclone-exposed coastal seas, and a natural design choice is to hand the network the cyclone as a prescribed input. We test that choice in the Bay of Bengal and find it harmful. We withhold 15 whole cyclones spanning 65 to 150 kt from GLORYS12 reanalysis and compare two U-Nets that are identical except for four prescribed cyclone-track channels. Across three seeds the ocean-only model beats persistence in every run and the storm-conditioned model loses to it in every run, with the two skill ranges disjoint (p = 3.1e-5, paired across storms). The cause is exposure frequency rather than signal content: the channels are non-zero on only 7.9% of training days, so they are out of distribution the moment they activate. The extra error falls inside the prescribed storm footprint, and replacing the real cyclone map with a no-storm map at inference improves held-out storm forecasts by 7.5 to 16.4% in every seed. The conditioned network has learned a response to a rare signal that is confidently wrong.
Graph coarsening reduces the large Quadratic Unconstrained Binary Optimization (QUBO) formulations arising when vehicle-routing problems are solved by quantum annealing. Nearby customers with compatible time windows are merged into super-nodes, the reduced problem is solved, and the solution is expanded to the original graph. For the Capacitated Vehicle Routing Problem with Time Windows (CVRPTW), existing coarsening heuristics require family-specific tuning and remain unreliable on random instances. We address these limitations on the Solomon benchmark using simulated annealing and a D-Wave Advantage2 processor.
We first introduce adaptive penalty calibration. Uniform penalty scaling has little effect, whereas controlling the internal coefficient range substantially improves raw samples. Removing non-binding constraints, normalising binding ones, and scaling the remaining penalties reduces mean raw constraint violations from 33.0 to 0.06 at the same solver budget (p=3.7e-11, n=56). A variable-count-preserving control attributes this gain to conditioning rather than problem size.
Second, we replace the hand-tuned merge score with a graph neural network (GNN) using one configuration across all families. At N=10, it achieves 100% feasibility across all Solomon families, including R-type (100% vs. 80% for the tuned heuristic). Across N=10,...,100, feasibility is 83% vs. 69%, with the GNN better or tied on 85/90 instance-size pairs. At N=80,100, the difference is significant (p=0.002; 25/25 pairs), while the QUBO remains approximately 5-6 times smaller.
Finally, hardware experiments reproduce the conditioning effect at fixed logical variable count: feasible samples increase from 0.02% to 39% across 13 instances. Classical repair with local search remains a reference bound for end-to-end solution cost.
The representation chosen for a mathematical operation can affect both its algebraic form and its empirical learning difficulty. We study this phenomenon for inversion over \(\mathbb F_{2^n}\), with field elements expressed in varying ordered \(\mathbb F_2\)-bases. We prove that two ordered bases induce the same coordinate inversion map if and only if they belong to the same Galois orbit. Since every orbit has size \(n\), the correspondence between ordered bases and distinct inversion maps is exactly \(n\)-to-one. We then analyze three Boolean formulations of inversion. The reference formulation has algebraic degree \(n-1\) and joint ANF leap \(1\), the mixed representation formulation has degree \(2(n-1)\) and joint ANF leap \(2\), and the complete raw formulation has degree at most \(3(n-1)\) and joint ANF leap at least \(n\). Exhaustive computations agree with the theoretical results and bounds in the cases considered. Controlled experiments with multilayer perceptrons show the same ordering in learning difficulty, while Galois orbit redundancy provides only a limited generalization benefit under the tested conditions. These results show that exact redundancy among representations can coexist with changes in Boolean structure and learning behavior when the representation is exposed as part of the input.
We investigate how optimizers scale across the overtraining axis and show that relative optimizer performance and optimal hyperparameters change substantially with training horizon. In particular, we study how matrix-preconditioned methods (Muon and SOAP) and a momentum-scheduled method (ADANA) scale relative to AdamW. We compare these four optimizers across models from 51M to 253M parameters and overtraining (OT) factors from 1x to 256x, sweeping the base learning rate at every setting. The preferred learning rate schedule can reverse across the overtraining axis, the best weight decay coefficient scales approximately as sqrt(OT), and longer horizons generally favor longer fixed memory. ADANA's scaling advantage over AdamW persists after tuning AdamW's fixed memory separately at each horizon. Log-time weight decay and momentum cooldown provide substantial gains for ADANA that compound as training increases. With this treatment, ADANA outscales AdamW with an exponent advantage close to that predicted by DANA theory on power-law random features. Muon and SOAP instead provide roughly constant token-efficiency advantages over AdamW across most of the measured range, although SOAP may gain further at the highest overtraining factors. ADANA begins behind both matrix-preconditioned optimizers but closes the gaps as training increases, surpassing Muon and becoming competitive with SOAP at our highest OT factors. These results establish training horizon as an essential axis for optimizer evaluation and design.
The FitzHugh-Nagumo (FHN) system serves as a simplified model of neuronal voltage dynamics, capturing the activator-inhibitor structure behind both isolated action potentials and the rhythmic spiking seen across the brain. Exploring its 5D physiological parameter space is important for neuromodulation and mapping voltage recordings back to biophysics, yet classical finite-difference solvers make rapid parameter sweeps expensive. We train parameter-conditioned Fourier Neural Operators (FNOs) as fast, differentiable surrogates for the FHN voltage and recovery fields on a one-dimensional spatial domain, conditioning each Fourier layer on the parameter vector $\lambda = (D_u, D_v, a, b, \tau)$ via feature-wise linear modulation (FiLM). We apply a single bifurcation analysis that delimits the two distinct regimes the model spans, oscillatory (tonic firing) and excitable (action-potential propagation), and we train one operator in each. In the oscillatory regime the surrogate attains sub-$0.1\%$ relative $L^2$ error on both fields, runs nearly three orders of magnitude faster than the finite-difference baseline, generalizes uniformly across the parameter space, and extrapolates to low single-digit percentage errors outside of the training bounds. In the excitable regime the same operator accurately reproduces the firing threshold and the $c \propto \sqrt{D_u}$ conduction-velocity law and replicates full traveling pulses, fully capturing the excitable bifurcation structure rather than just smoothly interpolating fields.
A weeklong summer workshop brought higher education faculty to campus to explore how AI and machine learning materials can be adapted for their classrooms.
A new machine-learning framework aims to improve the success rate of computational protein design while moving away from results that reproduce sequences found in nature.
MIT News · Artificial Intelligence· news.mit.eduAug 24, 2026
